genome_entropy

genome_entropy is a bioinformatics pipeline for measuring information content across representations derived from genomic DNA:

DNA → ORFs → proteins → 3Di + optional 12-state structural encodings → entropy

Current multitask ModernProst models emit both 3Di and 12-state (12st) encodings. Legacy ModernProst and ProstT5 models emit 3Di only. Raw Shannon entropy is calculated for available representations; normalised entropy is a downstream derived value and is not stored in standard JSON.

The pipeline calls the external get_orfs program for six-frame ORF discovery, uses pygenetic-code for translation, and supports FASTA or GenBank input. GenBank matching marks whether a called ORF corresponds to an annotated CDS; it does not assign biological function.

Start here

• Installation
  • Requirements
  • PyPI
  • Optional extras
  • Development checkout
  • External get_orfs dependency
  • CPU, NVIDIA, AMD, and Apple
  • Model downloads and offline use
  • Verification
• Quick start
  • Prerequisites
  • Complete pipeline
  • Step-by-step workflow
  • Protein-only workflow
  • Model and device selection
  • Next steps
• User guide
  • Pipeline concepts
  • ORFs and genetic codes
  • Structural-state encodings
  • Shannon entropy
  • Choosing an input workflow
  • Performance and reproducibility
  • Interpretation limits
• Models and structural-state encodings
  • Authoritative model registry
  • Output semantics
  • Loading, security, and cache behaviour
  • Devices, precision, and multi-GPU
  • Citations and provenance
• Data formats, coordinates, and entropy
  • Input formats
  • Unified pipeline JSON
  • Coordinates and identifiers
  • GenBank matching and in_genbank
  • Raw and normalised entropy
  • Intermediate and backward-compatible formats
• Machine-learning workflow
  • Purpose and interpretation
  • Models
  • Feature matrix and missing values
  • Splitting and reproducibility
  • Training, evaluation, and persistence
  • Prediction
  • Devices
  • Reference
• GPU and HPC operation
  • Platform semantics
  • Scheduler discovery and visibility
  • Installation and offline jobs
  • Monitoring
  • ML caveat

Reference

• Command-line reference
  • Global options
  • run
  • orf
  • translate
  • fasta-to-protein
  • encode3di
  • entropy
  • download
  • estimate-tokens
  • ml train
  • ml predict
  • Exit behaviour
  • Environment variables
• Python API reference
  • ORF discovery
  • Translation
  • Structural-state encoding
  • Entropy
  • Pipeline and schemas
  • I/O
  • Machine learning
  • Configuration, errors, and logging
• Encoding-budget estimation
  • CLI
  • Python API

Development

• Development guide
  • Environment
  • Source layout
  • Quality commands
  • Documentation
  • Integration tests
  • Release metadata
  • Contributing safely
• Changelog
  • Unreleased
  • 0.1.0 — 19 January 2026

Project status and support

The package is alpha software. The source repository, issue tracker, and licence are available at https://github.com/linsalrob/genome_entropy. Include the package version and accelerator details in bug reports, and do not publish credentials or sensitive sequence data.

Indices

• Index

• Module Index

• Search Page