Skip to main content
Ctrl+K

AlphaGenome PyTorch 0.3.2.dev17+g50ebc9821 documentation

Contents:

  • Installation
  • Quick Start
  • Named Outputs
  • Full Chromosome Prediction
  • Command-Line Interface (agt)
  • Performance Tips
  • Finetuning
    • Command-Line Interface
    • Python API
    • Adapters
    • API Reference
  • API Reference
    • Model
    • Layers
    • Training
  • .rst

AlphaGenome PyTorch Documentation

Contents

  • AlphaGenome PyTorch Documentation
  • Indices and tables

AlphaGenome PyTorch Documentation#

PyTorch implementation of the AlphaGenome DNA sequence-to-function model.

Contents:

  • Installation
    • Basic Installation
    • Building documentation
  • Quick Start
    • Loading the Model
    • Preparing Input
    • Inference
    • Understanding Outputs
    • GPU Inference
    • Mixed Precision
    • Next Steps
  • Named Outputs
    • Setup
    • Output types and track counts
    • Track metadata fields
    • Filtering tracks
    • Masks and indices
    • Resolution-independent queries
    • Cross-head filtering
    • Variant effect scoring
    • Padding tracks
    • Loading and building metadata catalogs
    • Comparison with JAX AlphaGenome
  • Full Chromosome Prediction
    • Command-Line Script
    • Python API
    • Tiling Configuration
    • Supported Heads
    • Performance Tips
    • API Reference
  • Command-Line Interface (agt)
    • Global options
    • agt info
    • agt predict
    • agt finetune
    • agt score
    • agt convert
    • agt preprocess
    • agt serve
    • Dependency Gating
  • Performance Tips
    • torch.compile
    • Mixed Precision
    • Resolution Selection
    • Head Selection
    • Gradient Checkpointing
    • Batch Size
    • Combining Optimizations
  • Finetuning
    • Overview
    • Quick Start
  • API Reference
    • Model
    • Layers
    • Training

Indices and tables#

  • Index

  • Module Index

  • Search Page

next

Installation

Contents
  • AlphaGenome PyTorch Documentation
  • Indices and tables

By Kundaje Lab

© Copyright 2026, Kundaje Lab.