Skip to main content

Chromatin features

Multi-label classification that predicts 919 mammalian chromatin features (DeepSEA-style) from DNA sequence. The default model (g0-deepsea) is a G0 BERT-Base encoder trained on ENCODE chromatin accessibility data across cell types and features.

Input​

A DNA sequence passed as sequence (inline) or a server-side handle. The model uses a 1,000 bp context window with a 200 bp prediction window.

Output​

Per-feature probabilities across 919 labels, grouped into eight feature categories (DNase, CTCF, Pol2, c-Myc, H3K27ac, H3K27me3, H3K4me1, and Other), returned in the {data, meta} envelope. Available output formats include probabilities, BED, and bigWig.

Try it​

REST API: POST /v1/tasks/chromatin/predict — see the getting-started guide. MCP: ask your client to fetch a sequence and predict its chromatin features. See mcp.md.