Can I use Geniml for single-cell ATAC-seq analysis?

Yes, Geniml includes the scEmbed module, which allows you to train cell-level embeddings on single-cell chromatin accessibility data for clustering and visualization.

What primary data formats does the Geniml skill support?

Geniml is specifically designed to work with genomic interval data, primarily in the form of BED files and single-cell formats like AnnData for chromatin accessibility datasets.

What is a 'universe' in the context of this genomic skill?

A universe is a reference set of consensus peaks created from multiple BED files. It serves as a standard coordinate system for tokenizing genomic regions during machine learning workflows.

Does Geniml require external machine learning libraries?

Yes, for machine learning capabilities like training embeddings, Geniml requires PyTorch. It also integrates with Scanpy for single-cell bioinformatics workflows.

Genomic Interval ML (Geniml)

Name: Genomic Interval ML (Geniml)
Author: ovachiever

byovachiever

•

Data Science & ML

Facilitates machine learning on genomic interval data, including embeddings for BED files and single-cell ATAC-seq analysis.

About

Geniml is a specialized skill designed for bioinformaticians and data scientists working with genomic region data. It enables unsupervised learning of embeddings for BED files, single cells, and metadata, providing a comprehensive toolkit for tasks like similarity searching, clustering, and building consensus peak universes. By integrating sophisticated methods like Region2Vec, BEDspace, and scEmbed, this skill streamlines the transition from raw genomic intervals to actionable machine learning models and downstream analysis within the Claude Code environment.

Key Features

Region2Vec for unsupervised genomic region embeddings and similarity analysis
BEDspace for joint embedding of genomic regions and experimental metadata labels
Genomic utilities for BED file caching, randomization, and embedding evaluation
scEmbed for single-cell ATAC-seq dimensionality reduction and cell-type clustering
Consensus peak 'universe' building using statistical methods like HMM and ML
15 GitHub stars

Use Cases

Performing metadata-aware similarity searches across large-scale genomic repositories
Training region embeddings for downstream genomic feature learning and predictive modeling
Analyzing and clustering single-cell chromatin accessibility data using Scanpy integration

About

Key Features

Region2Vec for unsupervised genomic region embeddings and similarity analysis
BEDspace for joint embedding of genomic regions and experimental metadata labels
Genomic utilities for BED file caching, randomization, and embedding evaluation
scEmbed for single-cell ATAC-seq dimensionality reduction and cell-type clustering
Consensus peak 'universe' building using statistical methods like HMM and ML
15 GitHub stars

Use Cases

Performing metadata-aware similarity searches across large-scale genomic repositories
Training region embeddings for downstream genomic feature learning and predictive modeling
Analyzing and clustering single-cell chromatin accessibility data using Scanpy integration