scvi-tools — анализ одноклеточных данных

★ 7.3 · content

scvi-tools is a Claude Code skill that enables deep learning-based single-cell analysis using the scvi-tools probabilistic modeling framework. It covers data integration and batch correction with scVI/scANVI, chromatin accessibility analysis with PeakVI, multi-modal CITE-seq analysis with totalVI, joint RNA+ATAC multiome analysis with MultiVI, spatial transcriptomics deconvolution with DestVI, label transfer and query-to-reference mapping via scANVI/scArches, and transcriptional dynamics with veloVI. The skill includes 12 reference files, a set of CLI scripts — prepare_data.py, train_model.py, cluster_embed.py, transfer_labels.py, and others — plus a model_utils.py utility library for building custom workflows. It is aimed at bioinformaticians working with variational autoencoders, multiome data, or latent space representations of single-cell genomics datasets.