scDeepCluster, is a single-cell model-based deep embedded clustering method, which simultaneously learns feature representation and clustering via explicit modelling of scRNA-seq data generation. Based on testing extensive simulated data and real datasets from four representative single-cell sequencing platforms, scDeepCluster outperformed state-of-the-art methods under various clustering performance metrics and exhibited improved scalability, with running time increasing linearly with sample size. Its accuracy and efficiency make scDeepCluster a promising algorithm for clustering large-scale scRNA-seq data.
Example use case: exploring cell heterogeneity and diversity in cancer research
Technology: Autoencoders
Prostate multiparametric MRI (mpMRI) studies comprising T2-weighted, Diffusion-weighted and T1-weighted contrast enhanced series.\u00a0
","mean_job_execution_time":{"hardware":"cpu","mean_job_execution_time_str":"9m \u00b1 5m"},"model_actions":[],"name":"Prostate Region Segmentation","run_count":17,"save_model":false,"status":"released","uid":105},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"627146f148139943a4273ae4\" # Brain Tumor Segmentation\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {}\n\nlocal_files = {\n \"input_file\": \"path/to/file_1.nii\" # allowed extensions: nii, gz\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"input_file\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"DKFZ","beautiful_url":null,"creation_time":"Jun 27 2022, 00:58:18","description":"Highlights tumor areas in brain MRI scans for easy review.","dev_version":1,"example_job_id":"6334bf93436e3cd135933d74","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Segment or classify medical images","group":"category","id":"67376a0e58410e1fd8b79b7b","name":"Image Analysis"}],"data_extension":[{"description":null,"group":"data_extension","id":"6322f3dc25be0d5520cf36de","name":"nii.gz"}],"focus":[{"description":"Revealing insights from visual data","group":"focus","id":"6436e6c2861ab51f63e18ba0","name":"Biomedical Image Analysis and Interpretation"}],"method":[{"description":null,"group":"method","id":"62d12325c119651f85f08562","name":"Image Segmentation"}],"organization":[{"description":null,"group":"organization","id":"67376662cd1f499d334d13b9","name":"DFKZ"}],"subject":[{"description":null,"group":"subject","id":"62d12215c119651f85f08553","name":"Anatomy"}]},"has_liked":false,"id":"627146f148139943a4273ae4","is_paid_app":false,"last_update_time":null,"like_count":10,"long_description":"Segmentation of possible glioma tumours and edemas on brain MRI scans using a pre-trained nnU-Net architecture. Given an MRI scan in .NIfTI format, this app will return your MRI scans with glioma and edema regions highlighted, available for download.
\nData requirements: Brain tumor MRI scans, with four MRI modalities as T1w, t1gd, T2w, and FLAIR.
\n","mean_job_execution_time":{"hardware":"gpu","mean_job_execution_time_str":"~1m"},"model_actions":[],"name":"Brain Tumor Segmentation","run_count":26,"save_model":false,"status":"released","uid":106},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"627320c048139943a4273ae5\" # Clipper\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"analysis\": \"differential\",\n \"index_column\": True\n}\n\nlocal_files = {\n \"control\": \"path/to/file_1.csv\", # allowed extensions: csv,\n \"experiment\": \"path/to/file_2.csv\" # allowed extensions: csv\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"control\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\"\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"Ge et al.","beautiful_url":null,"creation_time":"Jun 27 2022, 00:58:18","description":"Finds key features in scientific data efficiently.","dev_version":1,"example_job_id":"6335f399436e3cd135933d7f","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":null,"group":"organization","id":"6737666bcd1f499d334d13c0","name":"UCLA"}],"subject":[]},"has_liked":false,"id":"627320c048139943a4273ae5","is_paid_app":false,"last_update_time":null,"like_count":10,"long_description":"Identifying \"interesting\" features with false discovery rate (FDR) control where \"interesting\" means \"enriched\" or \"differential\", without using p-values
Example use case: identification of differentially expressed genes from genome-wide gene expression data or peak-calling from ChIP-seq data, peptide-identification from mass spectrometry data
Limitations: Works best with two conditions
scJoint leverages information from annotated scRNA-seq data in a semisupervised framework and uses a neural network to simultaneously train labeled and unlabeled data, allowing label transfer and joint visualization in an integrative framework. Using atlas data as well as multimodal datasets generated with ASAP-seq and CITE-seq, we demonstrate that scJoint is computationally efficient and consistently achieves substantially higher cell-type label accuracy than existing methods while providing meaningful joint visualizations. Thus, scJoint overcomes the heterogeneity of different data modalities to enable a more comprehensive understanding of cellular phenotypes.
","mean_job_execution_time":{},"model_actions":[],"name":"scJoint","run_count":24,"save_model":false,"status":"released","uid":114},{"added_by":"60f86f4d7f04e40b912f5b7a","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"62bb8db923181fc4793a92bd\" # Liver Cancer Segmentation\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {}\n\nlocal_files = {\n \"input_file\": \"path/to/file_1.gz\" # allowed extensions: gz\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"input_file\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"DKFZ","beautiful_url":null,"creation_time":"Jun 28 2022, 23:24:41","description":"Automatically identify liver and tumor in CT scans.","dev_version":1,"example_job_id":"6335f66ccec3574b2d34087a","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Segment or classify medical images","group":"category","id":"67376a0e58410e1fd8b79b7b","name":"Image Analysis"}],"data_extension":[{"description":null,"group":"data_extension","id":"6322f3dc25be0d5520cf36de","name":"nii.gz"}],"focus":[{"description":"Revealing insights from visual data","group":"focus","id":"6436e6c2861ab51f63e18ba0","name":"Biomedical Image Analysis and Interpretation"}],"method":[{"description":null,"group":"method","id":"62d12325c119651f85f08562","name":"Image Segmentation"}],"organization":[{"description":null,"group":"organization","id":"67376662cd1f499d334d13b9","name":"DFKZ"}],"subject":[{"description":null,"group":"subject","id":"62d12215c119651f85f08553","name":"Anatomy"}]},"has_liked":false,"id":"62bb8db923181fc4793a92bd","is_paid_app":false,"last_update_time":null,"like_count":5,"long_description":"Automatic segmentation of liver and liver tumor in contrast-enhanced abdominal CT scans using a pre-trained nnU-Net architecture.
\n","mean_job_execution_time":{"hardware":"gpu","mean_job_execution_time_str":"3m \u00b1 4m"},"model_actions":[],"name":"Liver Cancer Segmentation","run_count":7,"save_model":false,"status":"released","uid":115},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"62bcdc56dd9ecc514ae82877\" # Medical Image-nnUnet\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {}\n\nlocal_files = {\n \"input_file\": \"path/to/file_1.nii\" # allowed extensions: nii, gz\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"input_file\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\"\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"DKFZ","beautiful_url":null,"creation_time":"Jun 27 2022, 00:58:18","description":"Segments hippocampus into anterior and posterior regions","dev_version":1,"example_job_id":"6335f676cec3574b2d34087c","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Segment or classify medical images","group":"category","id":"67376a0e58410e1fd8b79b7b","name":"Image Analysis"}],"data_extension":[{"description":null,"group":"data_extension","id":"6322f3dc25be0d5520cf36de","name":"nii.gz"}],"focus":[{"description":"Revealing insights from visual data","group":"focus","id":"6436e6c2861ab51f63e18ba0","name":"Biomedical Image Analysis and Interpretation"}],"method":[{"description":null,"group":"method","id":"62d12325c119651f85f08562","name":"Image Segmentation"}],"organization":[{"description":null,"group":"organization","id":"67376662cd1f499d334d13b9","name":"DFKZ"}],"subject":[{"description":null,"group":"subject","id":"62d12215c119651f85f08553","name":"Anatomy"}]},"has_liked":false,"id":"62bcdc56dd9ecc514ae82877","is_paid_app":false,"last_update_time":null,"like_count":4,"long_description":"Segments the hippocampus into anterior and posterior regions
\n","mean_job_execution_time":{"hardware":"cpu","mean_job_execution_time_str":"~<1m"},"model_actions":[],"name":"Medical Image-nnUnet","run_count":2,"save_model":false,"status":"released","uid":116},{"added_by":"60f86f4d7f04e40b912f5b7a","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"62bf442025b09dead5853d24\" # AlphaFold2\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"aa_pairs\": \"[{'protein_name': 'PIAQIHILEGRSDEQKETLIREVSE'}]\",\n \"model_preset\": \"monomer\"\n}\n\nlocal_files = {}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"Jumper et al.","beautiful_url":null,"creation_time":"Jul 01 2022, 18:59:44","description":"Accurately predict protein structures for research and drug development.","dev_version":1,"example_job_id":"633acf45cec3574b2d3408a6","external_tool_library_id":null,"featured":null,"group":"alphafold","group_tags":{"category":[{"description":"Fold proteins, design antibodies","group":"category","id":"67376a1858410e1fd8b79b7d","name":"Protein Design"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":null,"group":"organization","id":"6737665ecd1f499d334d13b4","name":"DeepMind"}],"subject":[]},"has_liked":false,"id":"62bf442025b09dead5853d24","is_paid_app":false,"last_update_time":null,"like_count":24,"long_description":"Revolutionary 3D protein structure prediction model with high accuracy that allows for a close study of proteins without an expensive X-ray crystallography process
Example use case: Prediction of protein disorders, studying enzymes, and developing novel protein-based drugs
Technology: Novel neural network architectures and training procedures
Toxicity prediction for molecules in SMILES-format using ChemBERTa and the ClinTox dataset.
","mean_job_execution_time":{"hardware":"cpu","mean_job_execution_time_str":"~<1m"},"model_actions":[],"name":"Molecule Toxicity Prediction","run_count":30,"save_model":false,"status":"released","uid":119},{"added_by":"60f86f4d7f04e40b912f5b7a","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"62d06c17e35ad66addc86ad5\" # Pancreas Tumor Segmentation\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {}\n\nlocal_files = {\n \"input_file\": \"path/to/file_1.gz\" # allowed extensions: gz\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"input_file\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"DKFZ","beautiful_url":null,"creation_time":"Jul 14 2022, 19:18:47","description":"Automated tumor detection in pancreas CT scans","dev_version":1,"example_job_id":"6335f6d9cec3574b2d340882","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Segment or classify medical images","group":"category","id":"67376a0e58410e1fd8b79b7b","name":"Image Analysis"}],"data_extension":[{"description":null,"group":"data_extension","id":"6322f3dc25be0d5520cf36de","name":"nii.gz"}],"focus":[{"description":"Revealing insights from visual data","group":"focus","id":"6436e6c2861ab51f63e18ba0","name":"Biomedical Image Analysis and Interpretation"}],"method":[],"organization":[{"description":null,"group":"organization","id":"67376662cd1f499d334d13b9","name":"DFKZ"}],"subject":[]},"has_liked":false,"id":"62d06c17e35ad66addc86ad5","is_paid_app":false,"last_update_time":null,"like_count":4,"long_description":"Tumor identification and segmentation in pancreas based on the automated analysis of CT pancreas scans (3D volumes)
\n","mean_job_execution_time":{"hardware":"gpu","mean_job_execution_time_str":"1m \u00b1 1m"},"model_actions":[],"name":"Pancreas Tumor Segmentation","run_count":5,"save_model":false,"status":"released","uid":120},{"added_by":"60f86f4d7f04e40b912f5b7a","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"62d66daef86d46f6cf1e80a0\" # DeepPurpose-DTI\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"model\": \"Morgan_CNN_BindingDB\"\n}\n\nlocal_files = {\n \"target_file_input\": \"path/to/file_1.txt\", # allowed extensions: txt,\n \"drug_file_input\": \"path/to/file_2.tab\", # allowed extensions: tab,\n \"vs_file_input\": \"path/to/file_3.csv\" # allowed extensions: csv\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"target_file_input\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\"\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"Huang et al.","beautiful_url":null,"creation_time":"Jul 19 2022, 08:39:10","description":"Predicts drug and protein binding for drug discovery","dev_version":1,"example_job_id":"6335f715cec3574b2d340884","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Model docking, predict properties","group":"category","id":"673769f058410e1fd8b79b79","name":"Drug Design"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":null,"group":"organization","id":"67376671cd1f499d334d13c7","name":"Harvard"}],"subject":[]},"has_liked":false,"id":"62d66daef86d46f6cf1e80a0","is_paid_app":false,"last_update_time":null,"like_count":12,"long_description":"Measures the binding strength of drug molecules to the protein targets
Example use case: fundamental for drug discovery supporting drug screening and repurposing
Technology: Message passing neural network & Convolutional neural network
Detection and segmentation of cells and nuclei in microscopy. Works well with fluorescent images.
Example use case: image-based cellular research & disease
Technology: Convolutional neural network
Limitations: Currently only accepts black & white images
Nixtla's neuralforecast is a python library for time series forecasting with deep learning models. This app provides AutoRNN functionality, implemented using PyTorch, PyTorchLightning and Ray. This package automatically tests different settings, identifying over multiple iterations which parameters perform the best. The results for the champion model are then provided. If trained using CPU, then a maximum of 32 time series will be processed, while if trained using GPU a maximum of 128 will be processed instead. Models are trained using all series, meaning that with more series included, performance is likely to improve. Models can be saved, and then applied to similar time series, or trained further.
Please provide data in three columns, in the following order:
StatsForecast offers a collection of widely used univariate time series forecasting models, including automatic ARIMA and ETS modeling optimized for high performance using numba. It also includes a large battery of benchmarking models.
","mean_job_execution_time":{"hardware":"cpu","mean_job_execution_time_str":"2m \u00b1 4m"},"model_actions":[],"name":"ARIMA Time Series","run_count":2,"save_model":false,"status":"released","uid":132},{"added_by":"60f86f4d7f04e40b912f5b7a","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"62f361f9cfd06816d3a6eb72\" # ML Forecast\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"seasonality\": 1,\n \"horizon\": 10\n}\n\nlocal_files = {\n \"train_data\": \"path/to/file_1..csv\", # allowed extensions: .csv, .txt, .tsv,\n \"test_data\": \"path/to/file_2.csv\" # allowed extensions: csv, txt, tsv\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"train_data\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\"\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"Nixtla","beautiful_url":null,"creation_time":"Aug 10 2022, 07:44:57","description":"AI-driven time series predictions with adjustable seasonality.","dev_version":1,"example_job_id":"6336087fcec3574b2d340892","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Train AutoML models","group":"category","id":"67376a1658410e1fd8b79b7c","name":"Misc ML"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":"","group":"organization","id":"6737667bcd1f499d334d13d4","name":"NIXTLA"}],"subject":[]},"has_liked":false,"id":"62f361f9cfd06816d3a6eb72","is_paid_app":false,"last_update_time":null,"like_count":7,"long_description":"mlForecast allows users to apply scikit-learn models, such as random forests or gradient boosting to time series data. This particular implementation uses the hyperopt python package, automatically training numerous sklearn random forest, XGBoost, and LightGBM models, identifying over multiple iterations which parameters perform the best. The results for the champion model are then provided. If trained using CPU, then a maximum of 32 time series will be processed, while if trained using GPU a maximum of 128 will be processed instead, and the maximum iterations used to identify the best models will be higher. Models are trained using all series, meaning that with more series included performance is likely to improve. The seasonality parameter can be altered to reflect known cycles in the data (for example, if hourly data that is expected to vary in a regular pattern over a 24 hour period, then a seasonality of 24 may be suitable).A seasonality of 1 implies that no seasonality is included in the model.
Please provide data in three columns, in the following order:
Generates new molecules with desirable properties given a starting molecule, essentially replicating a chemist's intuition to optimize molecular properties, accelerating drug discovery
Example use case: molecular optimization for finding drug candidates
Limitations: works with 3 ADMET properties: logD, solubility and clearance
Technology: Message passing neural network
Metrics: As reported by He et al.
You can use GenNet to create neural networks specifically for genetics. GenNet gives you the ability to decide what knowledge should be connected to what.
\n","mean_job_execution_time":{},"model_actions":[],"name":"GenNet","run_count":17,"save_model":false,"status":"released","uid":136},{"added_by":"60f86f4d7f04e40b912f5b7a","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"63221a9e6a20e3ac924a7b80\" # DeepCRISPR: sgRNA Off-Targets\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"model\": \"Both\"\n}\n\nlocal_files = {\n \"input_file\": \"path/to/file_1.csv\" # allowed extensions: csv\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"input_file\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\"\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"Chuai et al.","beautiful_url":null,"creation_time":"Sep 14 2022, 18:17:02","description":"Deep learning tool to predict sgRNA off-target effects.","dev_version":1,"example_job_id":"63360908cec3574b2d340897","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":null,"group":"organization","id":"6737666fcd1f499d334d13c5","name":"Tongji University"}],"subject":[]},"has_liked":false,"id":"63221a9e6a20e3ac924a7b80","is_paid_app":false,"last_update_time":null,"like_count":12,"long_description":"An efficient and extendable computational model for prediction of CRISPR sgRNA off-target profile, facilitating optimized sgRNA design with high sensitivity and specificity
Example use case: Epigenetics research
Limitations: Different off-target assays may have different sensitivities
Technology: Convolutional neural network-based deep learning network, trained on two different cell types: 293-related cell lines (18 sgRNAs) and K562 t (12 sgRNAs)
Metrics: As reported by Chuai et al.
An efficient and extendable computational model for prediction of CRISPR sgRNA on-target knockout efficacy, facilitating optimized sgRNA with high sensitivity and specificity
Example use case: Epigenetics research
Limitations: Focuses on conventional NGG-based sgRNA design for SpCas9 in humans
Technology: Convolutional neural network-based deep learning network, trained on ~15,000 sgRNAs containing 1071 genes from
four different cell lines (hct116, hek293t, hela, and hl60) with redundancy removed and can generalise well in new cell types for
sgRNA on-target knockout efficacy prediction
Metrics: As reported by Chuai et al.
Trains an scVI variational auto encoder to identify differential expression in genes between observations. Data should be in .h5ad, .h5, or .csv format. Uses the Anndata python package, which natively supports .h5ad meaning that file format is most suitable.
\nExample use case: Differential expression analysis is one of the primary downstream analyses of single cell RNA-seq data to identify gene markers for cell type detection and also provide inputs to other secondary analyses.\u00a0
","mean_job_execution_time":{},"model_actions":[],"name":"scRNA-seq Differential Expression","run_count":19,"save_model":false,"status":"released","uid":141},{"added_by":"60f86f4d7f04e40b912f5b7a","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"63358e696b12ad0623bb6abb\" # ChemProp: Solubility\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {}\n\nlocal_files = {\n \"input_file\": \"path/to/file_1.csv\" # allowed extensions: csv\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"input_file\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\"\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"ChemProp","beautiful_url":null,"creation_time":"Sep 29 2022, 12:24:09","description":"App uses ChemProp machine learning library to calculate solubilities of a list of SMILES strings. The output is a .CSV file with the predicted log(S) solubility values.\n\nThe app was trained on the AqSolDB database of aqueous solubility.","dev_version":1,"example_job_id":"634030b0af675a84bede651c","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Model docking, predict properties","group":"category","id":"673769f058410e1fd8b79b79","name":"Drug Design"}],"data_extension":[{"description":null,"group":"data_extension","id":"6322f3c725be0d5520cf36dc","name":"csv"}],"focus":[{"description":"Unlocking insights from chemical data","group":"focus","id":"6436e69b861ab51f63e18b9b","name":"Cheminformatics"}],"method":[],"organization":[{"description":null,"group":"organization","id":"67376669cd1f499d334d13bf","name":"MIT"}],"subject":[{"description":null,"group":"subject","id":"62e047da0b2ea260941358ac","name":"Drug Discovery"}]},"has_liked":false,"id":"63358e696b12ad0623bb6abb","is_paid_app":false,"last_update_time":null,"like_count":9,"long_description":"Solubility model trained on AqSolDB using ChemProp
Example use case: Filtering out drug candidates based on solubility
Limitations: Applicable for organic molecules only, as the training dataset consisted only of organic molecules
Technology: Message passing neural networks
Metrics: On test data; Mean average error = 0.66; RMSE = 0.99; R2= 0.82
Classification model training for molecular property prediction by analyzing learned molecular representations
Example use case: drug toxicity prediction during lead generation
Technology: Message passing neural networks
scANVI model trained on C. Elegans dataset from: \u201cA lineage-resolved molecular atlas of C. elegans embryogenesis at single-cell resolution\u201d. Transcriptomes covering 89701 cells and 20222 genes
","mean_job_execution_time":{},"model_actions":[],"name":"C. Elegans Cell Annotation","run_count":1,"save_model":false,"status":"released","uid":150},{"added_by":"60f86f4d7f04e40b912f5b7a","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"633c3ecaeaafa791a6c10dff\" # PBMC Annotation\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"task_flag\": \"inference\",\n \"ann_epochs\": 100\n}\n\nlocal_files = {\n \"train_path\": \"path/to/file_1.csv\" # allowed extensions: csv, h5ad, h5\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"train_path\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"scVI-Tools","beautiful_url":null,"creation_time":"Oct 04 2022, 14:10:18","description":"Annotate and analyze PBMC dataset for single-cell studies.","dev_version":1,"example_job_id":null,"external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":null,"group":"organization","id":"67376682cd1f499d334d13db","name":"10X Genomics"}],"subject":[]},"has_liked":false,"id":"633c3ecaeaafa791a6c10dff","is_paid_app":false,"last_update_time":null,"like_count":7,"long_description":"scANVI model trained on Purified PBMC dataset from: \u201cMassively parallel digital transcriptional profiling of single cells\u201d. Transcriptomes covering 68k peripheral blood mononuclear cells.
","mean_job_execution_time":{},"model_actions":[],"name":"PBMC Annotation","run_count":8,"save_model":false,"status":"released","uid":151},{"added_by":"60f86f4d7f04e40b912f5b7a","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"633c3f59eaafa791a6c10e00\" # Mouse Spleen & Lymph Cell Annotation\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"task_flag\": \"inference\",\n \"ann_epochs\": 100\n}\n\nlocal_files = {\n \"train_path\": \"path/to/file_1.csv\" # allowed extensions: csv, h5ad, h5\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"train_path\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"scVI-Tools","beautiful_url":null,"creation_time":"Oct 04 2022, 14:12:41","description":"Labels immune cells in mouse spleen and lymph nodes.","dev_version":1,"example_job_id":null,"external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":null,"group":"organization","id":"6737667ccd1f499d334d13d5","name":"Weizmann Institute"}],"subject":[]},"has_liked":false,"id":"633c3f59eaafa791a6c10e00","is_paid_app":false,"last_update_time":null,"like_count":4,"long_description":"scANVI model trained on Mouse Spleen and Lymph Nodes Cite Seq dataset from: \u201cJoint probabilistic modeling of single-cell multi-omic data with totalVI\u201d. Immune cells from murine spleen and lymph nodes
","mean_job_execution_time":{},"model_actions":[],"name":"Mouse Spleen & Lymph Cell Annotation","run_count":1,"save_model":false,"status":"released","uid":152},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"633d7ed6f58e323c39c95df4\" # DRKG\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {}\n\nlocal_files = {\n \"heads\": \"path/to/file_1.csv\", # allowed extensions: csv,\n \"tails\": \"path/to/file_2.csv\", # allowed extensions: csv,\n \"relations\": \"path/to/file_3.csv\" # allowed extensions: csv\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"heads\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\"\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"Ioannidis et al.","beautiful_url":null,"creation_time":"Oct 05 2022, 12:55:50","description":"Links biological entities for drug discovery insights.","dev_version":1,"example_job_id":"633daf61f58e323c39c95e06","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Model docking, predict properties","group":"category","id":"673769f058410e1fd8b79b79","name":"Drug Design"}],"data_extension":[{"description":null,"group":"data_extension","id":"6322f3c725be0d5520cf36dc","name":"csv"}],"focus":[],"method":[{"description":null,"group":"method","id":"62e3a0fbb0daeec176142b2b","name":"Graph"}],"organization":[{"description":null,"group":"organization","id":"67376673cd1f499d334d13cb","name":"Amazon"}],"subject":[{"description":null,"group":"subject","id":"62e047da0b2ea260941358ac","name":"Drug Discovery"},{"description":null,"group":"subject","id":"62e3a0f3b0daeec176142996","name":"Drug"}]},"has_liked":false,"id":"633d7ed6f58e323c39c95df4","is_paid_app":false,"last_update_time":null,"like_count":4,"long_description":"Drug Repurposing Knowledge Graph (DRKG) is a comprehensive biological knowledge graph relating genes, compounds, diseases, biological processes, side effects and symptoms. DRKG includes information from six existing databases including DrugBank, Hetionet, GNBR, String, IntAct and DGIdb, and more. It includes 97,238 entities belonging to 13 entity-types; and 5,874,261 triplets belonging to 107 edge-types. These 107 edge-types show a type of interaction between one of the 17 entity-type pairs (multiple types of interactions are possible between the same entity-pair). See information on how to use DRKG here: https://bit.ly/3CcQCB5","mean_job_execution_time":{},"model_actions":[],"name":"DRKG","run_count":9,"save_model":false,"status":"released","uid":153},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"634419f6af675a84bede652b\" # Reinvent: DRD2 \n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"epochs\": 100,\n \"learning_rate\": 0.0001\n}\n\nlocal_files = {\n \"inception_smiles_file\": \"path/to/file_1.csv\", # allowed extensions: csv,\n \"avoid_smiles_file\": \"path/to/file_2.csv\" # allowed extensions: csv\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"inception_smiles_file\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\"\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"AstraZeneca","beautiful_url":null,"creation_time":"Oct 10 2022, 13:11:18","description":"This app explores new compounds for the DRD2 receptor by using reinforcement learning. ","dev_version":1,"example_job_id":"637df6812ffa50c6deecdab5","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Model docking, predict properties","group":"category","id":"673769f058410e1fd8b79b79","name":"Drug Design"}],"data_extension":[{"description":null,"group":"data_extension","id":"6322f3c725be0d5520cf36dc","name":"csv"}],"focus":[],"method":[{"description":null,"group":"method","id":"637de1732ffa50c6deecda83","name":"De novo Drug Design"},{"description":null,"group":"method","id":"62e3a0f3b0daeec1761429b1","name":"Reinforcement Learning"}],"organization":[{"description":null,"group":"organization","id":"67376684cd1f499d334d13dd","name":"AstraZeneca"}],"subject":[{"description":null,"group":"subject","id":"62e047da0b2ea260941358ac","name":"Drug Discovery"}]},"has_liked":false,"id":"634419f6af675a84bede652b","is_paid_app":false,"last_update_time":null,"like_count":5,"long_description":"Reinvent library consists of jupyter notebooks related to in silico drug discovery applications including data preparation and training steps. The aim of this app is to show the use case for finding new molecules for the DRD2 receptor with Reinforcement Learning. Inputs are two .csv files that SMILES strings should be provided. Please see the data structure below for more information. For more information, please see the tutorial page of Reinvent Apps.
Example Use Case: Finding new molecules for DRD2 receptor
Limitation: Some of the parameters are kept default like in the demo notebook.
Technology: Reinforcement Learning
Metrics: Some of the metrics shown in the demo notebook
This pipeline will train a UNet2D segmentation model that produces image masks that to represent the segmentations. The pipeline can augment the training dataset to increase model performance if you wish. To train a model you will need a set of source images and a set of target segmentation images. The target segmentation images will be binary images that are segmentation masks of the source images, each source image must have an associated target image. Model saving and GPU access available soon.
Example use case: Highlight blood vessels in a retinal image.
Technology: UNet2D is an encoder-decoder network architecture.
Limitations: The maximum image input size UNet2D can take is 512x512, your images will be converted to patches if they are bigger than 512x512. Currently only greyscale images are supported.
This pipeline will train a UNet2D multilabel segmentation model that produces image masks that represent the segmentations. The pipeline can augment the training dataset to increase model performance if you wish. To train a model you will need a set of source images and a set of target segmentation images. The target segmentation images will be that are segmentation masks of the source images where each pixel value is a label, each source image must have an associated target image. Model saving and GPU access available soon.
Example use case: Highlight various parts of cells
Technology: U-Net Encoder-decoder
Limitations: The maximum image input size UNet2D can take is 512x512, your images will be converted to patches if they are bigger than 512x512. Currently only greyscale images are supported
Train a model to denoise your medical images. Noise2Void is a deep-learning method that can be used to denoise many types of images, including microscopy images. It allows denoising of image data in a self-supervised manner, therefore high-quality, low noise equivalent images are not necessary to train this network. This is performed by \"masking\" a random subset of pixels in the noisy image and training the network to predict the values in these pixels. Model saving and GPU access available soon.
Example use case: Train a model to take noisey images and remove noise to make the image clearer without needing examples of the clear images to train the model
Technology: U-Net
Limitations: Currently only gray scale images are supported
This app provides how to process, analyze, and filter data from ChEMBL or other sources. There are several reasons why data used to train a generative model should be pre-processed. Invalid or duplicate entries need to be removed. Unusual compounds that are clearly not drug-like need to be excluded (too big, reactive groups and etc.) and rare tokens can be removed. There are some rare compounds that can be considered outliers and excluding them frees up space in the vocabulary, making it smaller. Input is SMILES strings in .smi format. Output is filtered SMILES in .csv format that can be used in other Reinvent apps. (e.g. Reinvent: Create Initial Prior/Agent Generative Model). For more information, please see the tutorial page of Reinvent Apps.
Example Use Case: Pre-processing SMILES strings from big sources for downstream analysis.
Limitation: Filtering options are kept like in the demo notebook. Other options will be added next versions.
Train a model to denoise your 3D medical images. Noise2Void is a deep-learning method that can be used to denoise many types of images, including microscopy images. It allows denoising of image data in a self-supervised manner, therefore high-quality, low noise equivalent images are not necessary to train this network. This is performed by \"masking\" a random subset of pixels in the noisy image and training the network to predict the values in these pixels. The resulting output is a denoised version of the image. Model saving and GPU access available soon.
Example use case: Train a model to take noisey images and remove noise to make the image clearer without needing examples of the clear images to train the model
Technology: U-Net
Ultrafast, precise, and parameter-free image reconstruction from high-density single-molecule localization microscopy (SMLM). Model saving and GPU access available soon.
Example use case: Research with single-molecule localization microscopy
Technology: U-Net based network without skip connections
Based on the Create_Model_Demo notebook from the ReinventCommunity repo. If you are planning to train a new model from scratch, this app is the first step of the process. This module creates an initial generative model of prior/agents for the next step (Reinvent: Train Initial Generative Model for Prior/Agent ) by generating a vocabulary based on your SMILES training data. The vocabulary defines what tokens the model can propose, which directly controls the possible atom types in the SMILES output. The output of this app (empty_model.ckpt) will be the input for the next app (Reinvent: Train Initial Generative Model for Prior/Agent ) that trains the generative model. Finally, the output file of the second app is what you can use in Reinvent: Reinforcement Learning. For more information, please see the tutorial page of Reinvent Apps.
Example Use Case: Generate generative model of prior/agents
Limitation: At least 100.000 SMILES are recommended for creating a generative model
DFCAN transforms low-resolution (LR) images to super-resolved (SR) images. The training is done using LR-SR image pairs, taking the LR images as input and obtaining an output as close to SR as possible. DFCAN achieves comparable image quality to SIM over a tenfold longer duration in multicolor live-cell imaging experiments, which reveal the detailed structures of mitochondrial cristae and nucleoids and the interaction dynamics of organelles and cytoskeleton. Model saving and GPU access available soon.
Example use case: Take low-resolution microscapy images and increase the resoution of them to reveal clearer structures in the images
Technology: Deep Fourier channel attention network\u00a0
This app was originally created based on notebook (Transfer Learning Mode) from the ReinventCommunity repo. The purpose of this app is to train an empty model created from the previous Reinvent: Create Initial prior/agent Generative Model. In this scenario, inputs are an empty model (.ckpt file) and a processed dataset (.csv file) generated by Reinvent: Data Preparation Module. The output will be the Prior that can be used afterward for\u00a0other apps of Reinvent such as reinforcement learning. For more information, please see the tutorial page of Reinvent Apps.
Example Use Case: Train generative model with Transfer Learning
Limitations: Some of the parameters are kept the same as in the original notebook.
Technology: Transfer Learning (TL)
Train pix2pix to translate one image domain to another image domain. The image transformation requires paired images for training (supervised learning) to learn information from the input image and obtain the equivalent translated image. Model saving and GPU access available soon.
Example use case: in silico cell painting, semantic segmentation, background removal, style transfer.
Technology: Generative Adversarial Network
This app derives from a notebook (Model Building Demo) in the ReinventCommunity repo. The aim of this app is to generate a QSAR model for a target protein that you are interested. QSAR modeling is an estimation of the activity/property/toxicity of new chemical entities in the drug discovery process. In the demo, SMILES that known activities on the DRD2 receptor were used. SMILES are transformed into fingerprints and act as the input in the code. Since REINVENT only supports pickled scikit-learn models, the output files are in (.pkl) format. The output file either can be used by (Reinvent: Reinforcement Learning Module) or (Reinvent: Scoring Module) apps. For more information, please see the tutorial page of Reinvent Apps.
Example Use Case: Generating QSAR models for target protein
Limitations:
- Only scikit-learn models (.pkl) are accepted by REINVENT
- Since the demo notebook contains only classification methodology for the random forest, the app provides this option currently.
Technology: Random Forest, scikit-learn, Quantitative structure-activity relationship
Metrics: AUC score metrics from demo data are represented here. Also, you can check the Example Output of this app.
Prediction of Tuberculosis diagnosis from chest X-ray images
\nExample use case: TB case detection
Tecnology: Deep convolutional neural network
Limitations: Do not use TB-Net for self-diagnosis and seek help from your local health authorities
Metrics: As reported by the TB-Net Open Source Initiative
Prediction of skin cancer diagnosis from dermoscopy images
\nExample use case: Skin cancer detection research
Tecnology: Deep convolutional neural network
Limitations: Do not use Cancer-Net SCa for self-diagnosis and seek help from your local health authorities
Metrics: As reported by the Cancer-Net SCa Open Source Initiative
CycleGAN is a method that can capture the characteristics of one image domain and learn how these characteristics can be translated into another image domain, all in the absence of any paired training examples. Model saving and GPU access available soon.
Example use case: In silico cell painting, semantic segmentation, background removal, style transfer.
Limitations: If your dataset is paired, use the pix2pix app instead, paired training generally provides more information to the deep learning model and so can perform more effectively.
Technology: Two Generative Adversairal Networks that learn to transform images both from the first domain to the second and vice-versa.
Image classification models have millions of parameters. Training them from scratch requires a lot of labelled training data and a lot of computing power. Transfer learning is a technique that shortcuts much of this by taking a piece of a model that has already been trained on a related task and reusing it in a new model. Model saving and GPU access available soon.
Example use case: Lesion, tumour, bacteria, animal etc. classification
Technology: Pretrained models from TensorFlow Hub.
Metrics: All models trained on the ImageNet dataset containing 14,197,122 images labelled under 1000 classes.
The reference notebook of this app is \"Reinforcement Learning Demo\" from the ReinventCommunity repository. The purpose of this app decides which molecules are \"good\" or \"bad\" by using RL. Generally, the generative model that was created before must be directed to a relevant region of chemical space that involve compounds of interest. To achieve this, a Reinforcement Learning methodology is used with several scoring functions defined by the user. During the RL, molecules are generated by the agent and receive a score between 0 and 1. For more information, please see the tutorial page of Reinvent Apps.
Example use case: Generate molecules with RL setup.
Limitations:
-Some of the parameters are kept as default and not shown here (they will be added next versions)
-Currently, only the CPU version is available
Technology: Reinforcement Learning
Metrics: See metrics in the original demo notebook this app is derived from
This app is derived from the \"Scoring Demo\" notebook in the ReinventCommunity repository. The purpose of this module is to estimate SMILES scores with the same scoring function as intended for the RL. For example, in the RL process inception, SMILES can be used. In order to evaluate whether these SMILES are good or not for your case independently, they can be used in this app to produce scores for evaluation. For more information, please see the tutorial page of Reinvent Apps.
Example use case: Evaluation of SMILES strings by utilizing the same scoring function in the RL
Limitation:
- Some of the parameters are kept as default and not shown here (they will be added next versions)
Technology: RDkit
This app is based on \"Sampling_Demo.ipynb\" from the ReinventCommunity notebook. The purpose of the notebook is to show how to sample further compounds from a given model state (agent). During the RL process, each iteration of the agent produces an output of a number of molecules and stores it inside. If more ideas are necessary for the project afterward, this can be achieved through this module. The sampling module enables you to extract molecules from the trained agent (given model state) to use for more investigation. For more information, please see the tutorial page of Reinvent Apps.
Example Use Case: A sampling of SMILES from the generative trained agent
Limitation:
- Job only can be submitted in CPU version
- Currently, maximum 5 million SMILES can be sampled
Data quality can be important for downstream analysis (e.g. alignment) and data cleaning is one of the most important steps to improve data quality. This app eliminates unnecessary reads from your sequence (e.g. adapters, primers, Poly-A tails). Inputs are FASTQ reads and adapters sequences. Output is an info table generated by the Cudatapt algorithm. Based on Cutadapt version v4.1.
Example Use Case: Remove unwanted 3' adapters from the FASTQ sequence
Limitation: Currently, only 3' adapters are accepted
A typical strategy for examining the relationship between genotype and phenotype while controlling for a number of factors is genome-wide association studies (GWAS). Covariates that could have non-linear or interaction effects are typically ignored by GWAS. When such non-linearity is detected, DeepNull can model the non-linear impact of variables on phenotypes. It learns this potentially complicated and non-linear relationship using a flexible deep neural network (DNN), and during association testing, makes adjustments for the network's expectation of phenotype (based on covariates only).
Example use case: Improve phenotype prediction and association power in GWAS
Technology: Deep Neural Network (DNN)
Limitation: Some of the DNN parameters were left as default. Please check this page for more information.
The approach to date has been to fit a model to a family of related sequences. The conventional setting is limited, since a new model must be trained for each prediction task. These state-of-the-art models use only zero-shot inference, without any supervision from experimental data or additional training, capturing the functional effects of sequence variation.
Example use case: Predicting protein sequence variants, along with the effects of these variants on protein function.
Technology: The training mode adjusts the number of models used to generate predictions. Fast uses a single model, standard three, and intensive five. Where multiple models are used their ensemble is used to provide overall metrics.\u00a0
ESM-2 outperforms all tested single-sequence protein language models across a range of structure prediction tasks. There are many ways to define a protein contact. Here we're using the definition of 8 angstroms between carbon beta atoms. Note that the position of the carbon beta is imputed from the position of the N, CA, and C atoms for each residue. Inputs required: 1-10 .a3m files (aligned fasta).
Example Use Case: predicting where amino acids at different points along a sequence are in close proximity. Contact maps are generated, which provide a 2D representation of the proximity of amino acids in 3D space. This can provide clues about protein function and 3D protein structure.
Technology: if CPU is selected when submitting the job, then the esm2_t12_35M_UR50D model will be used (35M parameters), while if GPU is selected, then the more powerful esm2_t33_650M_UR50D model will be used (650M parameters).\u00a0
ESM-2 outperforms all tested single-sequence protein language models across a range of structure prediction tasks. This app has three stages. First, an embedding (fixed-dimensional vector representation) is obtained for each mutated sequence. Second, PCA is applied to reduce the number of dimensions prior to modelling. Third, a regression model is trained that can predict the mutation \"effect\" score given the embedding. Three types of model will be compared: random forests, support vector machines, and K nearest neighbours.
Example Use Case: predicting the biological activity of mutations of a protein, using fixed embeddings from ESM. The output in this case describes scaled effect of the mutation.
Technology: if CPU is selected when submitting the job, then the esm2_t12_35M_UR50D model will be used (35M parameters), while if GPU is selected, then the more powerful esm2_t33_650M_UR50D model will be used (650M parameters).\u00a0
AutoZI is a deep generative model adapted from scVI allowing a gene-specific treatment of zero-inflation. Plots relating to batch effects and gene expression levels are also generated. Ideally, data should be provided in .h5ad format, which is native to the Anndata python package. Alternatively, .h5, .csv or.mtx files can be provided if their formatting is suitable.
Example use case: In single-cell RNA sequencing data, biological processes or technical factors may induce an overabundance of zero measurements. Existing probabilistic approaches to interpreting these data either model all genes as zero-inflated, or none. But the overabundance of zeros might be gene-specific. AutoZI can distinguish between zero-inflated and non-zero inflated genes.
LDVAE (Linearly-decoded Variational Auto-encoder, also called Linear scVI) is a flavor of scVI with a linear decoder.
The advantages of LDVAE are: (i) Can be used to interpret latent dimensions with factor loading matrix. (ii) Scalable to very large datasets (>1 million cells).
The limitations of LDVAE include: (i) Less capacity than scVI differential expression or scANVI for annotation, which use a neural network decoder. (ii) Less capable of integrating data with complex batch effects.\u00a0
CellAssign is a simple yet, efficient approach for annotating scRNA-seq data in the scenario in which cell-type-specific gene markers are known. The method also allows users to control for nuisance covariates like batch or donor. The scvi-tools implementation of CellAssign uses stochastic inference, such that CellAssign will scale to very large datasets.
\nThe advantages of CellAssign are: (i) Lightweight model that can be fit quickly, (ii) Ability to control for nuisance factors.
\nThe limitations of CellAssign include: (i) Requirement for a cell types by gene markers binary matrix, (ii) The simple linear model may not handle non-linear batch effects.
","mean_job_execution_time":{},"model_actions":[],"name":"Single Cell Assign: Annotation","run_count":4,"save_model":true,"status":"released","uid":195},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"63888da92ffa50c6deecdca2\" # GrapHiC\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"epoch\": 100,\n \"base\": \"GM12878-encode-0\",\n \"target\": \"GM12878-geo-raoetal\",\n \"epi_features\": \"All\",\n \"base_file_list\": \"GM12878-encode-0,GM12878-encode-1,GM12878-encode-2,GM12878-geo-026,GM12878-geo-033\",\n \"cell_line_name\": \"K562\",\n \"hic_data_resolution\": 10000,\n \"input_embedding_size\": 27\n}\n\nlocal_files = {\n \"hic_dataset_1\": \"path/to/file_1..tar.gz\", # allowed extensions: .tar.gz,\n \"epigenetic_dataset_1\": \"path/to/file_2.tar.gz\" # allowed extensions: tar.gz\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"hic_dataset_1\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\"\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"Murtaza et al.","beautiful_url":null,"creation_time":"Dec 01 2022, 11:19:05","description":"Easily complete missing genomic Hi-C data using graphs","dev_version":1,"example_job_id":"6399aa072ffa50c6deece0e3","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[],"focus":[{"description":"Interpreting biological data","group":"focus","id":"6436e6b8861ab51f63e18b9d","name":"Computational Biology and Bioinformatics"}],"method":[{"description":null,"group":"method","id":"6399eddf2ffa50c6deece103","name":"Graph Auto Encoder(GAE)"},{"description":null,"group":"method","id":"6399ed9f2ffa50c6deece0ff","name":"Hi-C"}],"organization":[{"description":null,"group":"organization","id":"67376676cd1f499d334d13ce","name":"Brown"}],"subject":[]},"has_liked":false,"id":"63888da92ffa50c6deecdca2","is_paid_app":false,"last_update_time":null,"like_count":0,"long_description":"The three-dimensional (3D) architecture of the genome at various structural scales can be comprehended and investigated by scientists using Hi-C, a high-throughput chromosomal conformation capture technique. Generally, Hi-C experiment data is stored in a contact map of size\u00a0N \u00d7 N. Each row and column correspond to fixed-width\u00a0N windows (\u201cbins\u201d) in the range of 1 kbps to 1 mbps. The examination of these contact maps revealed significant structural features such as topologically associated domains (TADs).
Building high-resolution Hi-C contact maps often necessitate billions of reads, which is often impossible. GrapHiC imputes Hi-C contact maps by reformulating the Hi-C data as a position-aware graph, using less expensive ChIP-seq signals, and proposing a generative graph-autoencoder that first encodes the input graph into a latent representation.
Example use case: Imputation of Hi-C data
Technology: Graph Auto-Encoder (GAE)
Limitation:
- Some of the parameters were left as default. Please check this page for more information.
Metrics: Detailed metrics of the study can be found in the supplementary file.
Epigenetic Features:
All: ['RAD-21', 'RNA-Pol2','CTCF', 'DNASE-Seq', 'H3K27ME3', 'H3K27AC', 'H3K36ME3', 'H3K4ME1', 'H3K4ME2', 'H3K4ME3', 'H3K79ME2', 'H3K9AC', 'H4K20ME1', 'H3K9ME3']
DNA-Acessibility: ['RAD-21', 'RNA-Pol2','CTCF', 'DNASE-Seq']
Repression-Marker: ['H3K27ME3', 'H3K4ME2', 'H4K20ME1']
Activating-Marker: ['H3K4ME3', 'H3K9AC', 'H3K9ME3']
Enchancer-Interaction-Marker: ['H3K36ME3', 'H3K79ME2']
Gene-Related: ['H3K36ME3', 'H3K79ME2']
HiC-Reg-Reduced: ['CTCF', 'DNASE-Seq', 'H4K20ME1', 'H3K27ME3', 'H3K9ME3', 'H3K9AC', 'H3K4ME1', 'H3K27AC']
High-grade serous carcinoma (HGSC) is a type of tumour that arises from the serous epithelial layer in the abdominopelvic cavity and is mainly found in the ovary. This pretrained model can be used to generate cell type predictions, based on precalculated gene-cell associations. Note that the marker genes must be present in the uploaded dataset.\u00a0
","mean_job_execution_time":{},"model_actions":[],"name":"Single Cell Assign: HGSC","run_count":2,"save_model":false,"status":"released","uid":197},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"63889f942ffa50c6deecdcbe\" # Single Cell Assign: Lymphoma\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"epochs\": 400,\n \"batch_effect\": None\n}\n\nlocal_files = {\n \"train_path\": \"path/to/file_1.csv\" # allowed extensions: csv, h5ad, h5\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"train_path\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"scVI-Tools","beautiful_url":null,"creation_time":"Dec 01 2022, 12:35:32","description":"A pretrained model for generating follicular lymphoma cell type predictions","dev_version":1,"example_job_id":null,"external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":null,"group":"organization","id":"67376688cd1f499d334d13de","name":"BC Cancer Research Centre"}],"subject":[]},"has_liked":false,"id":"63889f942ffa50c6deecdcbe","is_paid_app":false,"last_update_time":null,"like_count":0,"long_description":"Follicular lymphoma is the most common type of of low-grade non-Hodgkin lymphoma (NHL). It develops when white blood cells cluster together to form lumps in lymph glands or organs. There are two main types of lymphoma: Hodgkin lymphoma and non-Hodgkin lymphoma. This pretrained model can be used to generate cell type predictions, based on precalculated gene-cell associations. Note that the marker genes must be present in the uploaded dataset.
","mean_job_execution_time":{"hardware":"cpu","mean_job_execution_time_str":"1m \u00b1 1m"},"model_actions":[],"name":"Single Cell Assign: Lymphoma","run_count":0,"save_model":false,"status":"released","uid":198},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"638fa7232ffa50c6deecde34\" # Stereoscope for Spatial Omics\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"target\": \"cell_states\",\n \"top_n\": 6000,\n \"sc_epochs\": 100,\n \"st_epochs\": 2000,\n \"batch_effect\": None\n}\n\nlocal_files = {\n \"sc_path\": \"path/to/file_1.csv\", # allowed extensions: csv, mtx, h5ad, h5,\n \"st_path\": \"path/to/file_2.csv\", # allowed extensions: csv, mtx, h5ad, h5,\n \"tissue_path\": \"path/to/file_3.csv\" # allowed extensions: csv, tsv, txt\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"sc_path\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"scVI-Tools","beautiful_url":null,"creation_time":"Dec 06 2022, 20:33:39","description":"Spatial gene activity mapping using RNA sequencing data","dev_version":1,"example_job_id":"63909add2ffa50c6deecde7c","external_tool_library_id":null,"featured":false,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":null,"group":"organization","id":"67376677cd1f499d334d13cf","name":"Sanger Institute"}],"subject":[]},"has_liked":false,"id":"638fa7232ffa50c6deecde34","is_paid_app":false,"last_update_time":null,"like_count":4,"long_description":"Stereoscope posits a probabilistic model of spatial transcriptomics and an associated method for the deconvoluton of cell type profiles using a single-cell RNA sequencing reference dataset.
The advantages of Stereoscope are: (i) Can stratify cells into discrete cell types, (ii) Scalable to very large datasets (>1 million cells).
Stereoscope requires training two latent variable models (LVMs): one for the single-cell reference dataset and one for the spatial transcriptomics dataset, which incorporates the learned parameters of the single-cell reference LVM. The first LVM takes in as input a scRNA-seq gene expression matrix of UMI counts with cells and genes, along with a vector of cell type labels. Subsequently, the second LVM takes in the learned parameters of the first LVM, along with a spatial gene expression matrix with spots and genes. This matrix should be uploaded as a csv file with gene names, 3D location, and tissue values in different columns.
Example use case: Train a probabilistic model of spatial transcriptomics (measure all the gene activity in a tissue sample and map where the activity is occurring) and an associated method for the deconvoluton of cell type profiles (estimating the proportions of different cell types in samples collected from a tissue) using a single-cell RNA sequencing reference dataset.
Limitations: Effectively requires a GPU for fast inference.
SynthSeg is the first convolutional neural network for segmentation of brain scans of any contrast and resolution that works out-of-the-box, without retraining or fine-tuning. SynthSeg relies on a single model, which is robust to a wide array of subject populations: from young and healthy to ageing and diseased subjects, white matter lesions, scans with or without preprocessing, including bias field corruption, skull stripping, intensity normalisation, template registration, etc.
Example use case: SynthSeg can segment real scans of any target domain, which enables analysis of huge amounts of heterogeneous clinical data.
Technology: Convoluional neural network
Metrics: Trained on a synthetic dataset. Obtains a dice score of > 0.85 on most availiable brain MRI datasets
totalVI (total Variational Inference) provides a flexible generative model of CITE-seq RNA and protein data that can be used for many common downstream tasks.
The advantages of totalVI are: (i) Comprehensive in capabilities. (ii) Scalable to very large datasets (>1 million cells).
Data should include the protein expression matrix, with one row per observation, and one column per protein.
Example use case: Train a generative model of CITE-seq RNA and protein data that can be used for many common downstream tasks.
Limitations: Effectively requires a GPU for fast inference. And, difficult to understand the balance between RNA and protein data in the low-dimensional representation of cells.
DestVI (Deconvolution of Spatial Transcriptomics profiles using Variational Inference) posits a conditional generative model of spatial transcriptomics down to the sub-cell-type variation level which can be used to explore the spatial organization of a tissue and understanding gene expression variation between tissues and conditions.
The advantages of DestVI are: (i) Can stratify cells into discrete cell types and model continuous sub-cell-type variation, (ii) Scalable to very large datasets (>1 million cells).
Example use case: Deconvolution of, for example, 10x Visium spatial transcriptomics profiles using an accompanying single-cell RNA sequencing data.
Limitations: Effectively requires a GPU for fast inference.
Structure-based drug design (SBDD) aims to design small-molecule ligands that bind with high affinity and specificity to pre-determined protein targets. This app generates small-molecule ligands that bind with high affinity and specificity to pre-determined protein targets using AI diffusion.
Example use case: Generate ligands for a binding pocket inside a target of interest to elcit a desired therapeutic effect
Technology: E(3)-equivariant 3D-conditional diffusion
One of the most significant biological challenges in the post-genomic era is understanding the functional roles and investigating the mechanisms of newly found proteins. A deep learning technique called DeepFRI uses both sequences and contact map representations of 3D structures to predict protein function. The protein structures from the PDB and SWISS-MODEL are used to train DeepFRI. LSTM-LM(Long Short-Term Memory Language Model) was used to learn features from protein sequences and GCN was used to discover features from contact maps.
Example use case: Predicting protein functions
Technology: Graph Convolutional Networks (GCN), LSTM-LM
Limitation: Some of the options to predict protein functions are currently not available. Please check this page for more information.
Metrics: Some of the metrics related to work can be found here
Tangram is for mapping single-cell (or single-nucleus) gene expression data onto spatial gene expression data. The single-cell dataset and the spatial dataset should be collected from the same anatomical region/tissue type, ideally from a biological replicate, and need to share a set of genes. Tangram aligns the single-cell data in space by fitting gene expression on the shared genes. Spatial data need to be organized as a voxel-by-gene matrix. The voxel coordinates are saved in the fields obs.x and obs.y which we can use to visualize the spatial ROI. Each \"dot\" is the center of a 10um voxel.
Example use case: The most common application of Tangram is to resolve cell types in space. Another usage is to correct gene expression from spatial data: as scRNA-seq data are less prone to dropout than (e.g.) Visium or Slide-seq, the \"new\" spatial data generated by Tangram resolve many more genes. As a result, we can visualize program usage in space, which can be used for ligand-receptor pair discovery or, more generally, cell-cell communication mechanisms. If cell segmentation is available, Tangram can be also used for deconvolution of spatial data. If your single cell are multimodal, Tangram can be used to spatially resolve other modalities, such as chromatin accessibility.
Superbio AutoML: Classification Workflow provides an intuitive, no-code approach to building machine learning models for classifying data into distinct categories. Whether working with binary or multiclass problems, this workflow leverages robust algorithms like XGBoost, ensemble methods, and bagging to identify the best-performing model tailored to your dataset.
\nUsers can effortlessly upload CSV data, define their target variable, and select evaluation metrics such as the Matthews Correlation Coefficient (MCC) for binary classification. The platform ensures transparency by allowing users to set the test data split percentage, making it easy to gauge the model's real-world performance.
\nTo achieve optimal results, the workflow emphasizes the importance of clean, leakage-free datasets. With a focus on simplifying the complexities of machine learning, this AutoML workflow empowers users to develop powerful classification models efficiently, regardless of their technical expertise.
\n
Limitations: Data should not have leakage and be as clean as possible.
Technology: Classical ML models including XgBoost, Ensamble, and Bagging.
Use cases: Create a classification model (binary or multiclass) from tabular data.
Superbio AutoML: Regression Workflow offers an intuitive, no-code platform for building machine learning models to predict continuous outcomes from tabular data. Leveraging powerful algorithms like XGBoost, ensemble methods, and bagging, this workflow automatically identifies the best-performing model for your dataset.
\nUsers can easily upload their CSV data, specify the target variable, and choose evaluation metrics such as Root Mean Square Error (RMSE) to measure the model\u2019s accuracy. The platform provides flexibility by allowing users to set the test data split percentage, ensuring clear insights into the model's ability to generalize to new, unseen data.
\nTo achieve the most reliable predictions, the workflow highlights the importance of clean, leakage-free datasets. By simplifying the complexities of regression modeling, this AutoML application empowers users to develop and deploy predictive models efficiently, regardless of their experience in data science.
\n
Limitations: Data should not have leakage and be as clean as possible.
Technology: Classical ML models including XgBoost, Ensamble, and Bagging.
Use cases: Create a regression model from tabular data.
Predicts binding and inhibition of molecules, in SMILES format, to human beta-secretace1 (BACE-1). BACE-1 has been implicated as a central player in the pathogenesis of Alzheimer's disease. Our model was fine-tuned with the BACE dataset and uses the ChemBERTa transformer.
Example use case: Alzheimer's research.
Technology: ChemBERTa Transformer of 77 million molecules.
Metrics: Test-AUC 0.845
Predicts possible undesirable side effects on the skin and subcutaneous tissue of molecules (SMILES format). Our model was fine-tuned with the SIDER dataset and uses the ChemBERTa transformer. We follow the standard \"System Organ Classes\" nomenclature.
Example use case: Drug interactions and side effects
Technology: ChemBERTa Transformer of 77 million molecules.
Metrics: Test-AUC 0.73
Pre-mRNAs are spliced into mature transcripts with extraordinary precision, but it is still unclear how the cellular machinery manages to be so particular. SpliceAI is a deep residual neural network that just requires the genomic sequence of the pre-mRNA transcript as input to predict whether each point in a pre-mRNA transcript is a splice donor, acceptor, or neither.
Example use case: Predicting splice variants
Technology: Deep Neural Network (DNN)
Limitation: Some of the parameters are kept as default. Please check this page for more information.
Metrics: Some of the metrics related to work can be found in the article.
Predicts possible undesirable product issues of molecules (SMILES format). Our model was fine-tuned with the SIDER dataset and uses the ChemBERTa transformer. We follow the standard \"System Organ Classes\" nomenclature.
Example use case: Drug interactions and side effects.
Technology: ChemBERTa Transformer of 77 million molecules.
Metrics: Test-AUC 0.66
Predicts possible Investigations issues of molecules (SMILES format). The issues include laboratory tests and other medical investigations that gave an unusual reading. Our model was fine-tuned with the SIDER dataset and uses the ChemBERTa transformer. We follow the standard \"System Organ Classes\" nomenclature.
Example use case: Drug interactions and side effects
Technology: ChemBERTa Transformer of 77 million molecules.
Metrics: Test-AUC 0.67
Predicts possible undesirable General disorders and administration site conditions of molecules (SMILES format). These are a class of disorders that encompasses conditions of a general kind that result from a disease, the treatment of disease or administration of treatment at a particular site and are manifested by a characteristic set of symptoms and signs. Our model was fine-tuned with the SIDER dataset and uses the ChemBERTa transformer. We follow the standard \"System Organ Classes\" nomenclature.
Example use case: Drug interactions and side effects.
Technology: ChemBERTa Transformer of 77 million molecules.
Metrics: Test-AUC 0.71
Predicts if a molecule may produce undesirable eye disorders. Our model was fine-tuned with the SIDER dataset and uses the ChemBERTa transformer. We follow the standard \"System Organ Classes\" nomenclature.
Example use case: Drug interactions and side effects
Technology: ChemBERTa Transformer of 77 million molecules.
Metrics: Test-AUC 0.65
Predicts if a molecule may cause Endocrine disorders. Our model was fine-tuned with the SIDER dataset and uses the ChemBERTa transformer. We follow the standard \"System Organ Classes\" nomenclature.
Example use case: Drug interactions and side effects.
Technology: ChemBERTa Transformer of 77 million molecules.
Metrics
Test-AUC 0.67
\n","mean_job_execution_time":{"hardware":"cpu","mean_job_execution_time_str":"~<1m"},"model_actions":[],"name":"ChemBERTa: Endocrine","run_count":1,"save_model":false,"status":"released","uid":222},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"639bcf6d2ffa50c6deece33a\" # ChemBERTa: Blood\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"smiles_format\": \"canonical_smiles\"\n}\n\nlocal_files = {\n \"inference_data\": \"path/to/file_1.csv\" # allowed extensions: csv\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"inference_data\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\"\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"Superbio.ai & ChemBERTa","beautiful_url":null,"creation_time":"Dec 16 2022, 01:52:45","description":"Predicts blood-related side effects of molecules.","dev_version":1,"example_job_id":null,"external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Model docking, predict properties","group":"category","id":"673769f058410e1fd8b79b79","name":"Drug Design"}],"data_extension":[{"description":null,"group":"data_extension","id":"6322f3c725be0d5520cf36dc","name":"csv"}],"focus":[{"description":"Unlocking insights from chemical data","group":"focus","id":"6436e69b861ab51f63e18b9b","name":"Cheminformatics"}],"method":[{"description":null,"group":"method","id":"62e3a0f2b0daeec176142946","name":"Classification"}],"organization":[{"description":null,"group":"organization","id":"67376665cd1f499d334d13be","name":"DeepChem"}],"subject":[{"description":null,"group":"subject","id":"639aaf632ffa50c6deece26e","name":"Drug Side Effects"}]},"has_liked":false,"id":"639bcf6d2ffa50c6deece33a","is_paid_app":false,"last_update_time":null,"like_count":1,"long_description":"Predicts if a molecule may cause Blood and lymphatic system disorders. Our model was fine-tuned with the SIDER dataset and uses the ChemBERTa transformer. We follow the standard \"System Organ Classes\" nomenclature.
\nExample use case: Drug interactions and side effects.
Technology: ChemBERTa Transformer of 77 million molecules.
Metrics: Test-AUC 0.67
Predicts if a molecule may cause Musculoskeletal and connective tissue disorders. Our model was fine-tuned with the SIDER dataset and uses the ChemBERTa transformer. We follow the standard \"System Organ Classes\" nomenclature.
Example use case: Drug interactions and side effects.
Technology: ChemBERTa Transformer of 77 million molecules.
Metrics: Test-AUC 0.72
Predicts if a molecule may cause Gastrointestinal disorders. Our model was fine-tuned with the SIDER dataset and uses the ChemBERTa transformer. We follow the standard \"System Organ Classes\" nomenclature.
Example use case: Drug interactions and side effects.
Technology: ChemBERTa Transformer of 77 million molecules.
Metrics: Test-AUC 0.78
Predicts if a molecule may cause Renal and urinary disorders. Our model was fine-tuned with the SIDER dataset and uses the ChemBERTa transformer. We follow the standard \"System Organ Classes\" nomenclature.
Example use case: Drug interactions and side effects.
Technology: ChemBERTa Transformer of 77 million molecules.
Metrics: Test-AUC 0.69
Predicts if a molecule may cause Respiratory, thoracic and mediastinal disorders. Our model was fine-tuned with the SIDER dataset and uses the ChemBERTa transformer. We follow the standard \"System Organ Classes\" nomenclature.
Example use case: Drug interactions and side effects.
Technology: ChemBERTa Transformer of 77 million molecules.
Metrics
Test-AUC 0.67
Predicts if a molecule may contribute to Infections and infestations caused by pathogens. Our model was fine-tuned with the SIDER dataset and uses the ChemBERTa transformer. We follow the standard \"System Organ Classes\" nomenclature.
Example use case: Drug interactions and side effects.
Technology: ChemBERTa Transformer of 77 million molecules.
Metrics: Test-AUC 0.67
DeepConsensus (v1.2) uses gap-aware sequence transformers to correct errors in Pacific Biosciences (PacBio) Circular Consensus Sequencing (CCS) data.\u00a0This results in greater yield of high-quality reads.
Example use case: Correct errors in Pacific Biosciences (PacBio) Circular Consensus Sequencing (CCS) data.
Technology: Gap-aware sequence transformers
Limitation: This is not an official Google product. The content of this research code repository (i) is not intended to be a medical device; and (ii) is not intended for clinical use of any kind, including but not limited to diagnosis or prognosis.
Metrics: Some of the metrics related to work can be found here
Computational programs speed up the processes of chemical discovery, but they frequently require accurate three-dimensional molecular data as part of the input. Obtaining optimal molecular structures is difficult because it necessitates the enumeration and optimization of a vast space of stereoisomers and conformers. Auto3D allows you to obtain favorable 3D conformations of organic molecules from an automatically generated stereoisomeric conformational space that has been optimized by atomistic neural network potentials (NNPs) like ANI or AIMNet.
Example use case: Generating 3D structures from SMILES
Technology: Neural Network Potentials (NNPs)
Limitation:
- The current approach does not take into account the physical environment, and all calculations were done in the gas phase.
- In real-world applications, some chiral molecules are used in racemic form or have poorly defined stereo centers, so assigning specific stereo information may result in over-curation.
- The current NNPs of Auto3D optimize geometries in a vacuum condition.
- Only AIMNET and ANI2x models are currently available.
- Currently, twenty 3D structures of SMILES can be generated.
Metrics: Some of the metrics related to work can be found in the article.
peakVI (Python class PEAKVI) is a generative model of scATAC-seq data that can subsequently be used for many common downstream tasks. The advantages of peakVI are: (i) Comprehensive in capabilities. (ii) Scalable to very large datasets (>1 million cells).
Example use case: Analyze and visualize scATACseq data
Limitations:
-Effectively requires a GPU for fast inference.
-Latent space is not interpretable, unlike that of a linear method.
MultiVI multimodal generative model capable of integrating multiome, scRNA-seq and scATAC-seq data. After training, it can be used for many common downstream tasks, and also for imputation of a missing modality.
The advantages of multiVI are: (i) Comprehensive in capabilities. Able to perform DE gene, DA region analysis. (ii) Scalable to very large datasets (>1 million cells). (iii) Once trained with sufficient multimodal data, able to accurately input missing modalities.
Important: MultiVI requires the datasets to use shared features. scATAC-seq datasets need to be processed to use a shared set of peaks. This features dataset must be included in the inputs
Example use case: MultiVI can be used for many common downstream tasks, and also for imputation of a missing modality
Limitations: Effectively requires a GPU for fast inference.
DiffDock-L (v.1.1.3) is the latest and improved version of DiffDock, a cutting-edge method for predicting molecular docking and drug binding structures. DiffDock-L offers a major enhancement in both performance and generalization capabilities, making it a more advanced and reliable tool for drug discovery and related applications.
\nExample use case: Predict binding structure of a ligand to a protein for the sake of drug discovery
\nTechnology: Diffusion
\nLimitation:
\nMetrics:
\npeakVI (Python class PEAKVI) is a generative model of scATAC-seq data that can subsequently be used for many common downstream tasks. Here we present a pretrained PBMC (white blood cell) dataset which can be used for differential accessibility analysis.
\n","mean_job_execution_time":{},"model_actions":[],"name":"PeakVI for PBMC","run_count":0,"save_model":false,"status":"released","uid":235},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"63dd1aecbd2a3db57fdf1e42\" # FoldingDiff\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"num\": 10,\n \"length_1\": 50,\n \"length_2\": 128\n}\n\nlocal_files = {}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"Microsoft Research","beautiful_url":null,"creation_time":"Feb 03 2023, 14:32:12","description":"Generates 3D protein structures for biological research.","dev_version":1,"example_job_id":"63dd1d8ebd2a3db57fdf1e56","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Fold proteins, design antibodies","group":"category","id":"67376a1858410e1fd8b79b7d","name":"Protein Design"}],"data_extension":[],"focus":[{"description":"Understanding complex biomolecular structures","group":"focus","id":"6436e68c861ab51f63e18b99","name":"Structural Bioinformatics"}],"method":[{"description":null,"group":"method","id":"62d122fac119651f85f0855d","name":"3D Protein Structure Prediction"}],"organization":[{"description":null,"group":"organization","id":"6737665fcd1f499d334d13b6","name":"Microsoft"}],"subject":[]},"has_liked":false,"id":"63dd1aecbd2a3db57fdf1e42","is_paid_app":false,"last_update_time":null,"like_count":5,"long_description":"Since they are involved in practically every biological activity, proteins are essential for life. The capacity to create innovative, physically foldable protein structures using computation may result in new biological understandings and therapeutic developments for diseases without cures. In this study, inter-residue angles in protein backbones are used rather than cartesian atom coordinates. Training a denoising diffusion probabilistic model with a basic transformer backbone shows that the model produces very realistic protein structures with complexity and structural patterns that are similar to those of naturally occurring proteins.
\nExample Use Case: Generating 3D protein backbone structures\u00a0from the pre-trained model.
\nTechnology: Diffusion Model
\nLimitations:
\nMetrics: Some metrics\u00a0related to TM-Scores can be found in the article.
","mean_job_execution_time":{},"model_actions":[],"name":"FoldingDiff","run_count":44,"save_model":false,"status":"released","uid":240},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"640b4aa1b7dfb773cbe47e15\" # DeepVariant\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"reference_genome\": \"ensembl_grch37\",\n \"model_type\": \"WES\"\n}\n\nlocal_files = {\n \"input_file\": \"path/to/file_1.bam\", # allowed extensions: bam, bam.bai,\n \"bai_file\": \"path/to/file_2.bai\" # allowed extensions: bai\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"input_file\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"Google Research","beautiful_url":null,"creation_time":"Mar 10 2023, 15:20:01","description":"AI app detects genetic variations for health insights.","dev_version":1,"example_job_id":null,"external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":null,"group":"organization","id":"6737666dcd1f499d334d13c2","name":"Google"}],"subject":[]},"has_liked":false,"id":"640b4aa1b7dfb773cbe47e15","is_paid_app":false,"last_update_time":null,"like_count":1,"long_description":"DeepVariant (v1.5) is a deep learning-based variant caller that analyzes and identifies variants associated with a specific trait, disease, or population.
\nExample use case: Disease research and precision medicine
\nTechnology: Convolutional neural network
\nMetrics: As reported by the Genomics team in Google Health, \"DeepVariant maintains high accuracy across data from different sequencing technologies, prep methods, and species.\"
","mean_job_execution_time":{},"model_actions":[],"name":"DeepVariant","run_count":3,"save_model":false,"status":"released","uid":249},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"641db05c9d35843a448c20fd\" # BOLT-LMM\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"phenoCol\": \"Your parameter value here\",\n \"covarCol\": \"Your parameter value here\",\n \"qCovarCol\": \"Your parameter value here\",\n \"impute2MinMAF\": 0,\n \"Nautosomes\": 22,\n \"verboseStats\": False,\n \"geneticMapFile\": None\n}\n\nlocal_files = {\n \"fam\": \"path/to/file_1.fam\", # allowed extensions: fam, fam.gz,\n \"bim\": \"path/to/file_2.bim\", # allowed extensions: bim, bim.gz,\n \"bed\": \"path/to/file_3.bed\", # allowed extensions: bed, bed.gz,\n \"phenoFile\": \"path/to/file_4.tab\", # allowed extensions: tab, tab.gz, txt, txt.gz,\n \"LDscoresFile\": \"path/to/file_5.tab.gz\", # allowed extensions: tab.gz,\n \"remove\": \"path/to/file_6.txt\", # allowed extensions: txt,\n \"exclude\": \"path/to/file_7.txt\", # allowed extensions: txt, txt.gz,\n \"covarFile\": \"path/to/file_8.tab\", # allowed extensions: tab, tab.gz, txt, txt.gz,\n \"modelSnps\": \"path/to/file_9.txt\", # allowed extensions: txt,\n \"dosageFile\": \"path/to/file_10.txt\", # allowed extensions: txt, tab, gz,\n \"dosageFidIidFile\": \"path/to/file_11.txt\", # allowed extensions: txt, txt.gz,\n \"impute2files\": \"path/to/file_12.txt\", # allowed extensions: txt, gz, tab,\n \"impute2FileList\": \"path/to/file_13.txt\", # allowed extensions: txt, gz,\n \"impute2FidIidFile\": \"path/to/file_14.txt\", # allowed extensions: txt, gz,\n \"2dosagefiles\": \"path/to/file_15.txt\", # allowed extensions: txt, gz,\n \"2dosagemapfile\": \"path/to/file_16.map\", # allowed extensions: map,\n \"dosage2FileList\": \"path/to/file_17.txt\" # allowed extensions: txt\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"fam\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\"\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"Loh et al.","beautiful_url":null,"creation_time":"Mar 24 2023, 14:14:52","description":"Analyzes genetic traits and variants using advanced statistical models.","dev_version":1,"example_job_id":"642a960200e71434164ca8ec","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[{"description":null,"group":"data_extension","id":"6322f41425be0d5520cf36e6","name":"bed"},{"description":null,"group":"data_extension","id":"6322f40e25be0d5520cf36e5","name":"bim"},{"description":null,"group":"data_extension","id":"6322f41a25be0d5520cf36e7","name":"fam"}],"focus":[{"description":"Understanding complex biomolecular structures","group":"focus","id":"6436e68c861ab51f63e18b99","name":"Structural Bioinformatics"}],"method":[{"description":null,"group":"method","id":"6380c3022ffa50c6deecdb30","name":"GWAS"}],"organization":[{"description":null,"group":"organization","id":"67376661cd1f499d334d13b7","name":"Broad Institute"}],"subject":[]},"has_liked":false,"id":"641db05c9d35843a448c20fd","is_paid_app":false,"last_update_time":null,"like_count":1,"long_description":"The BOLT-LMM (v2.4.1) algorithm employs a linear mixed model (LMM) to calculate statistical measures for examining the relationship between a phenotype (observable trait) and genotypes (genetic information). BOLT-LMM assumes a Bayesian mixture of normals before the random impact attributed to SNPs other than the one being tested by default. This model generalizes the traditional \"infinitesimal\" mixed model employed by prior mixed-model association approaches (e.g., EMMAX, FaST-LMM, GEMMA, GRAMMAR-Gamma, GCTA-LOCO), allowing for enhanced detection power while reducing false positives.
Example use case: GWAS (Genome-Wide Association Study)
Technology: Linear mixed model
Limitations:
- Currently, bgen format option is not available.
- BOLT-LMM is recommended for analyses of human genetic datasets with more than 5,000 samples.
- It is also noted that association test statistics obtained from BOLT-LMM are valid for quantitative traits as well as (reasonably) balanced case-control traits.
- The BOLT-LMM method, similar to other mixed-model approaches, can experience reduced effectiveness when applied to the analysis of large sets of case-control data in rare diseases, which may result in decreased statistical power.
- The research conducted does not aim to determine how much population structure or relatedness may affect the heritability parameter (h2g) estimated by BOLT-LMM, nor does it carry out or assess genetic prediction using external validation samples from a separate group.
- The performance of mixed-model techniques has not been examined in datasets where family structure plays a significant role.
- BOLT-LMM has only been evaluated on datasets consisting of human genetic data, which exhibit distinct genetic architectures and patterns of linkage disequilibrium compared to plant and animal data.
Metrics: Some of the metrics related to the study can be found in the article.
Transcription factors play a critical role in connecting the DNA sequence to gene expression in different types of cells. By understanding how, where, and when these factors bind to chromatin, we can gain insights into the gene regulatory networks that control cellular behavior. This knowledge can help us unravel the complex mechanisms behind gene expression, potentially leading to new therapies for various diseases. maxATAC is the most extensive collection of state-of-the-art TFBS models available to date, with models available for 127 human TFs. Additionally, MaxATAC can perform well on primary cells and single-cell ATAC-seq data, making it a valuable tool for predicting TFBS in vivo.
\nExample use case: Predicting TFs' binding locations can significantly aid biological research by supplying reference materials for experimental validation.
\nTechnology: Deep Neural Network
\nLimitation:
\nA deep learning-based basecaller by Oxford Nanopore Technologies (ONT)
\nExample use case: Sequencing
\nTechnology: A neural network architecture that consists of a single convolutional layer followed by three stacked bidirectional gated recurrent unit layers
\nLimitations: Works only with reads from ONT devices - does not work well with modified mRNA reads
","mean_job_execution_time":{},"model_actions":[],"name":"Bonito","run_count":1,"save_model":false,"status":"released","uid":258},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"646f2d37a81ca5a299cbecb7\" # Nextflow DIA MS Pipeline\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"msstats_normalization\": \"equalizeMedians\",\n \"aggregate\": False,\n \"msstats_reports\": False\n}\n\nlocal_files = {\n \"input_csv\": \"path/to/file_1.csv\", # allowed extensions: csv,\n \"data_files\": \"path/to/file_2.raw\", # allowed extensions: raw, mzML.gz,\n \"dlib_file\": \"path/to/file_3.dlib\", # allowed extensions: dlib,\n \"fasta_file\": \"path/to/file_4.fasta\", # allowed extensions: fasta,\n \"contrast_file\": \"path/to/file_5.csv\" # allowed extensions: csv\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"input_csv\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\"\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"Allen et al.","beautiful_url":null,"creation_time":"May 25 2023, 09:41:11","description":"Analyze DIA proteomics with seamless pipeline integration","dev_version":1,"example_job_id":"646f83b9a81ca5a299cbed03","external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[{"description":null,"group":"data_extension","id":"6322f3c725be0d5520cf36dc","name":"csv"},{"description":null,"group":"data_extension","id":"6322f4f025be0d5520cf36eb","name":"fasta"}],"focus":[{"description":"Unveiling the intricate world of proteins","group":"focus","id":"646f2c37a81ca5a299cbecb5","name":"Proteomics"}],"method":[{"description":"","group":"method","id":"646f2c4aa81ca5a299cbecb6","name":"Mass Spectrometry"}],"organization":[{"description":null,"group":"organization","id":"6737667ecd1f499d334d13d6","name":"Talus Bio"}],"subject":[]},"has_liked":false,"id":"646f2d37a81ca5a299cbecb7","is_paid_app":false,"last_update_time":null,"like_count":2,"long_description":"Data-independent acquisition (DIA) mass spectrometry techniques are highly valuable for comprehensive and systematic quantification of the proteome. However, the current landscape of open-source tools dedicated to analyzing DIA proteomics experiments remains limited. Moreover, the effective utilization of gas phase fractionated (GPF) chromatogram libraries for improved peptide detection and quantification in DIA experiments is lacking.
\nTo bridge this gap, a novel open-source NextFlow pipeline called nf-encyclopedia was developed. This pipeline seamlessly integrates three open-source tools, namely MSConvert, EncyclopeDIA, and MSstats, forming a comprehensive solution for analyzing DIA proteomics experiments, regardless of the use of chromatogram libraries. Further information on this app can be found here.
\nExample use case: DIA mass spectrometry data analysis
\nTechnology: Nextflow
\nLimitation: Some of the parameters were kept default. Please see this page for more detail.
","mean_job_execution_time":{},"model_actions":[],"name":"Nextflow DIA MS Pipeline","run_count":1,"save_model":false,"status":"released","uid":259},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"6489ea4aa81ca5a299cbf02b\" # scRNA-Seq Annotation: Custom Model\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"target\": \"cell_type\",\n \"top_n\": 1500,\n \"scvi_epochs\": 400,\n \"ann_epochs\": 25,\n \"downsample_ratio\": 0.2,\n \"n_latent\": 8,\n \"n_hidden\": 64,\n \"n_layers\": 2,\n \"dropout_rate\": 0.1\n}\n\nlocal_files = {\n \"train_path\": \"path/to/file_1.csv\" # allowed extensions: csv, h5ad, h5\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"train_path\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"scVI-Tools","beautiful_url":null,"creation_time":"Jun 14 2023, 16:26:50","description":"Custom model for single-cell RNA data annotation.","dev_version":1,"example_job_id":null,"external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":null,"group":"organization","id":"67376672cd1f499d334d13c9","name":"Berkeley"}],"subject":[]},"has_liked":false,"id":"6489ea4aa81ca5a299cbf02b","is_paid_app":false,"last_update_time":null,"like_count":1,"long_description":"Train a customized single cell annotation model, using scVI-Tools. Data should be in .h5ad, .h5, mtx, or .csv format. Uses the Anndata python package, which natively supports .h5ad meaning that file format is most suitable. The target should be the feature you want to label: most commonly this will be cell type. Please ensure that the column with these values is selected. A model will be trained to predict cell type from input data, which can then be saved and used for other data sets, even where cell type is not known.
Example use case: Train a model that can annotate cell types after clustering single-cell RNA sequencing (scRNA-seq) data.
scANVI model trained on Heart Atlas dataset from The Heart Cell Atlas Database. Combined single cell and single nuclei RNA-Seq data of 485K cardiac cells with annotations.
","mean_job_execution_time":{},"model_actions":[],"name":"Heart Cell Annotation","run_count":5,"save_model":false,"status":"released","uid":262},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"6489ebe0a81ca5a299cbf036\" # Liver Cell Annotation\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"task_flag\": \"inference\",\n \"ann_epochs\": 100\n}\n\nlocal_files = {\n \"train_path\": \"path/to/file_1.csv\" # allowed extensions: csv, mtx, h5ad, h5\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"train_path\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"scVI-Tools","beautiful_url":null,"creation_time":"Jun 14 2023, 16:33:36","description":"Analyzes liver cell data to identify fibrosis markers.","dev_version":1,"example_job_id":null,"external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":null,"group":"organization","id":"67376679cd1f499d334d13d1","name":"Novartis"}],"subject":[]},"has_liked":false,"id":"6489ebe0a81ca5a299cbf036","is_paid_app":false,"last_update_time":null,"like_count":2,"long_description":"scANVI model trained on human liver cell dataset from: \"Single-cell and bulk transcriptomics of the liver reveals potential targets of NASH with fibrosis.\"
","mean_job_execution_time":{},"model_actions":[],"name":"Liver Cell Annotation","run_count":3,"save_model":false,"status":"released","uid":263},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"6489ec86a81ca5a299cbf03b\" # Mouse Retina Cell Annotation\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"task_flag\": \"inference\",\n \"ann_epochs\": 100\n}\n\nlocal_files = {\n \"train_path\": \"path/to/file_1.csv\" # allowed extensions: csv, h5ad, h5\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"train_path\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"scVI-Tools","beautiful_url":null,"creation_time":"Jun 14 2023, 16:36:22","description":"Annotate mouse retina cells using single-cell data.","dev_version":1,"example_job_id":null,"external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":null,"group":"organization","id":"6737667ccd1f499d334d13d5","name":"Weizmann Institute"}],"subject":[]},"has_liked":false,"id":"6489ec86a81ca5a299cbf03b","is_paid_app":false,"last_update_time":null,"like_count":1,"long_description":"scANVI model trained on Mouse retinal bipolar neuron drop seq dataset from: \"Comprehensive classification of retinal bipolar neurons by single-cell transcriptomics.\"
","mean_job_execution_time":{},"model_actions":[],"name":"Mouse Retina Cell Annotation","run_count":2,"save_model":false,"status":"released","uid":264},{"added_by":"618232acba0c04f899e6f740","api_example":"# pip install superbio\n\nfrom superbio import Client\n\nemail = \"your_email\"\npassword = \"YOUR_PASSWORD\" # Replace with your actual password\nclient = Client(email, password)\n\napp_id = \"6489ed39a81ca5a299cbf040\" # scRNA-Seq Annotation: Auto Train\n\n# Note: Adjust running mode, config values, and file paths as needed.\njob_config = {\n \"target\": \"cell_type\",\n \"mode\": \"Fast\"\n}\n\nlocal_files = {\n \"train_path\": \"path/to/file_1.csv\" # allowed extensions: csv, h5ad, h5\n}\n\n# Alternatively, use remote_file_source_data for S3 files:\nremote_file_source_data = {\n \"train_path\": [\n {\n \"protocol\": \"s3\",\n \"credentials\": {\n \"aws_access_key_id\": \"YOUR_ACCESS_KEY\",\n \"aws_secret_access_key\": \"YOUR_SECRET_KEY\"\n },\n \"path\": \"your-bucket/path/to/remote/file.csv\"\n }\n ],\n # ...\n}\n\nrunning_mode = \"cpu\" # or gpu\n\njob = client.post_job(\n app_id=app_id,\n running_mode=running_mode,\n config=job_config,\n local_files=local_files,\n # remote_file_source_data=remote_file_source_data\n)\n\n# Wait for job to finish...\n\n# Download all results\njob_id = job[\"job_id\"]\nclient.download_all_job_results(\n job_id=job_id,\n path_to_download_to=\"path/to/save/results/in/\" # Local directory to save files\n)\n","author":"scVI-Tools","beautiful_url":null,"creation_time":"Jun 14 2023, 16:39:21","description":"Train models to annotate single cell RNA-Seq data.","dev_version":1,"example_job_id":null,"external_tool_library_id":null,"featured":null,"group":null,"group_tags":{"category":[{"description":"Run omics analysis","group":"category","id":"673769fa58410e1fd8b79b7a","name":"Omics"}],"data_extension":[],"focus":[],"method":[],"organization":[{"description":null,"group":"organization","id":"67376672cd1f499d334d13c9","name":"Berkeley"}],"subject":[]},"has_liked":false,"id":"6489ed39a81ca5a299cbf040","is_paid_app":false,"last_update_time":null,"like_count":3,"long_description":"Automatically train a single cell annotation model, using scVI-Tools. Data should be in .h5ad, .h5, mtx, or .csv format. Uses the Anndata python package, which natively supports .h5ad meaning that file format is most suitable. The target should be the feature you want to label: most commonly this will be cell type. Please ensure that the column with these values is selected. A model will be trained to predict cell type from input data, which can then be saved and used for other data sets, even where cell type is not known.
Example use case: Train a model that can annotate cell types after clustering single-cell RNA sequencing (scRNA-seq) data
Cell mapping consortia have generated large-scale single-cell datasets to understand cellular processes. However, analyzing datasets from different species together remains challenging. SATURN is a deep learning approach that integrates cross-species single-cell RNA expression data by combining gene expression with protein embeddings from language models. SATURN introduces the concept of macrogenes, functionally related gene groups, based on protein representations.
\nIt enables multi-species differential expression analysis and facilitates the mapping of datasets to a universal embedding space. SATURN is applied to integrate diverse datasets like the Human Cell Atlas, Mouse Cell Atlas, and Fly Cell Atlas, resulting in a comprehensive mammalian cell atlas. It also integrates frog and zebrafish embryogenesis datasets.
\nSATURN successfully transfers annotations across species, identifies homologous and species-specific cell types, and reveals potential functional differences in glaucoma-associated genes. By leveraging protein language models, SATURN improves cross-species analysis, enhancing our understanding of evolutionary processes and the link between cellular diversity and anatomical/physiological variation.
\nExample use case: Performing multi-species single-cell differential expression analysis.
\nTechnology: Embeddings, Large Protein Language Models
\nLimitation:
\nSimulation of molecular behavior is crucial in the field of new substance/drug research. Density functional theory (DFT) is frequently utilized when it's necessary to accurately account for quantum effects. However, it calls for a lot of computer power, particularly for large molecules. Simulating a large number of molecular structures is challenging. The package makes it simple to use deep learning on molecular structures. For instance, you can use machine learning to classify toxicity or to do HOMO (highest occupied molecular orbital) level regression given compound input.
\nExample use case: Chemical property prediction tool with GCNN
\nTechnology: Graph Convolutional Neural Network (GCNN)
","mean_job_execution_time":{"hardware":"gpu","mean_job_execution_time_str":"1h25m \u00b1 6h14m"},"model_actions":[],"name":"Chainer Chemistry","run_count":6,"save_model":false,"status":"released","uid":269}],"hits_per_page":10,"total":669}