borzoi

What problem does it solve?

The borzoi Skill addresses the need for predicting genome-wide functional tracks from DNA sequences, enabling users to score regulatory effects on expression/accessibility, generate coverage tracks for loci, and prioritize non-coding variants by predicted track delta.

Core Features & Use Cases

• Functional Track Prediction: Uses the Borzoi model to predict tracks (RNA-seq, CAGE, DNase, ChIP) from DNA sequences.
• Variant Scoring: Assess the regulatory impact of variants on expression or accessibility.
• Coverage Prediction: Generate predicted coverage tracks for a specific locus.
• Variant Prioritization: Prioritize non-coding variants based on predicted track changes.
• Use Case: A researcher could use borzoi to predict the impact of a variant on gene expression before conducting further experiments.

Quick Start

Run the 'borzoi' skill with the command 'predict_tracks(dna_sequence)'. Ensure you have a CUDA-enabled Python environment with Python 3.11 or higher, and CUDA 12.1 or later.