Models SDK (Coming soon...)

delong-models is proposed as an open-source framework designed to unify the fragmented landscape of Computational Biology and Bioinformatics. It aims to provide a standardized API for defining, loading, and training machine learning models across diverse biological modalities (Genomics, Proteomics, Cheminformatics).

Primary Objectives

  • Unification: Create a single API surface to handle models currently scattered across disparate and heterogeneous models & algorithms.

  • Abstraction: Remove the complexity of preprocessing by automating the conversion of biological entities into model-ready tensors.

  • Compliance: Embed legal and ethical metadata (Licensing, Data Provenance) directly into the model configuration to ensure enterprise viability.

  • Performance: Natively support High-Performance Computing (HPC) techniques (Flash-Attention, Sequence Packing) for Large Biological Models (LBMs).

Last updated