How does Geniml handle single-cell ATAC-seq data?

It uses the scEmbed capability to train Region2Vec models on single-cell chromatin accessibility data, generating cell-level embeddings that can be used for clustering and cell-type annotation.

What are 'universes' in genomic machine learning?

Universes are reference peak sets or consensus regions built from multiple BED files. They act as a standardized coordinate system for tokenizing genomic data before training ML models.

What is Geniml in the context of Claude Code?

Geniml is a Python library and Claude Code skill that provides unsupervised machine learning methods for genomic interval data, such as BED files, allowing AI to assist in complex biological data analysis.

Can I perform cross-modal searches with this skill?

Yes, using the BEDspace module, you can train joint embeddings for both genomic regions and metadata labels, enabling queries like finding regions associated with specific experimental conditions.

Geniml: Genomic Machine Learning

Name: Geniml: Genomic Machine Learning
Author: jimmc414

byjimmc414

•

324

•

Data Science & ML

Applies unsupervised machine learning models to genomic interval data for region embeddings, single-cell analysis, and similarity searches.

Geniml is a specialized Claude Code skill designed for bioinformaticians and researchers to perform advanced machine learning on genomic interval data (BED files). It streamlines complex workflows like training Region2Vec for region embeddings, scEmbed for single-cell ATAC-seq analysis, and building consensus peak sets (universes). By integrating Geniml, users can automate the generation of embeddings for similarity analysis, cross-modal metadata queries, and downstream genomic feature learning directly within their AI-driven scientific discovery environment, bridging the gap between raw genomic data and actionable machine learning insights.

Key Features

01Region2Vec for unsupervised genomic region embeddings and dimensionality reduction

02scEmbed for cell-level embeddings and clustering of single-cell ATAC-seq data

03Automated 'universe' building for creating statistically rigorous consensus peak sets

04BEDspace for joint region and metadata embeddings to enable cross-modal queries

05Integration with the scanpy ecosystem and BEDbase for streamlined bioinformatics workflows

06324 GitHub stars

Use Cases

01Building searchable genomic databases using joint region-metadata embeddings for experimental discovery

02Standardizing heterogeneous BED file collections into consensus references for tokenization

03Dimensionality reduction and clustering of single-cell chromatin accessibility datasets

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jimmc414/kosmos geniml

For use in Claude.ai and ChatGPT

Download Skill