Skip to content

Latest commit

 

History

History
34 lines (24 loc) · 1.11 KB

File metadata and controls

34 lines (24 loc) · 1.11 KB

SeqTrainer Vision

SeqTrainer aims to become a "Keras for synthetic biology" in spirit: easy domain APIs, composable defaults, and clear extension points.

Core principle

SeqTrainer should own the domain layer:

  • SynBioHub/SBOL retrieval and interoperability
  • SPARQL recipes for common synbio data patterns
  • Dataset recipe abstractions and provenance
  • DNA-aware transforms and feature extraction
  • Task-focused applications (e.g., promoter regression)

SeqTrainer should not replace model ecosystems:

  • Keras remains a high-level training interface
  • PyTorch remains native for HF/DNABERT and graph ecosystems

Architecture direction

  1. Framework-neutral core first
  2. Optional framework adapters
  3. Backbone + head composition
  4. Pluggable application recipes
  5. Strong provenance and reproducibility story over time

Near-term roadmap

  • Expand SynBioHub client auth + pagination support
  • Curate reusable SPARQL recipe library for SBOL entities
  • Add robust dataset caching/snapshot materialization
  • Add production-grade torch/keras model factories
  • Promote graph prototype scripts into stable graph APIs