
scBERT is a pretrained deep language model for annotating cell types in single-cell RNA-seq data. It is pretrained on unlabeled data and then fine-tuned on your labeled data.
Web appFree
Step 1: Download the pretrained checkpoint and the data from the links in the README.
Step 2: Pre-process your data: revise gene symbols and normalize with scanpy.
Step 3: Run the fine-tuning script with your data path and the pretrained model path.
Step 4: Run the prediction script with the fine-tuned model to get a cell type for each cell.
Step 5: Turn on the novel cell type flag, with a custom threshold if needed, to mark cells of new types.
By thresholding predicted probabilities. The default threshold is 0.5 and you can set your own.
Revise gene symbols against the NCBI Gene database, which removes unmatched and duplicate genes, then normalize with scanpy.
About 30 minutes to install and about 25 minutes to infer 10,000 cells on a normal desktop.
No. The README says it is for research and not approved for clinical use.
napari AI ecosystemnapari is a fast, interactive viewer for multi-dimensional images in Python. It displays arrays, overlays points, polygons and segmentations, and lets you annotate and edit them.
ImmuneBuilderImmuneBuilder is a set of deep learning models that predict the structures of antibodies, nanobodies and T-cell receptors from sequence. It is installed with pip and used from Python or the command line.
DeepChemDeepChem is an open-source Python library for machine learning and deep learning on molecular and quantum datasets. It builds on PyTorch and other popular frameworks.