Geniml — ML-анализ геномных интервалов в формате BED

★ 7.0 · ai-tooling

geniml is a Claude Code skill that orchestrates machine-learning and statistical workflows over genomic interval sets stored as BED files. It validates BED structure and universe contracts against local chromosome-sizes files, plans Region2Vec and scEmbed training runs, inspects model and tokenizer compatibility, and assesses consensus universes — all locally, with no network requests made by the bundled helpers. The skill targets geniml 0.8.4 with gtars 0.9.2 on Python 3.10+; ML workflows additionally require the pinned ml extra for Torch, Gensim, Scanpy, and Hugging Face Hub dependencies. Bioinformaticians analyzing scATAC-seq chromatin accessibility, building genomic-region embeddings, or managing multi-sample BED corpora will find explicit handling of assembly contracts, coordinate conventions, and sensitive sample metadata built into every planning step.