README.md

metadata

license: other
license_name: cc-by-nc-4.0
license_link: https://creativecommons.org/licenses/by-nc/4.0/legalcode
tags:
  - enzyme
  - protein
  - biology
  - ec-prediction
  - multimodal
  - contrastive-learning
language:
  - en
pipeline_tag: feature-extraction

RAMER

This Hugging Face repository stores the official resources for RAMER (reaction-aware multimodal enzyme function representation model).

What is stored in this repository

The repository mainly includes three resource groups:

• model/
  Model weights and tokenizer/config files required for RAMER inference and training reproduction.

• data/
  Benchmark and evaluation data (for example, test CSV/JSON files and related resources used in EC prediction workflows).

• Background_library/
  Background embedding/index resources and dictionary files used by zero-shot retrieval pipelines (e.g., EC label dictionaries and background H5 files).

Intended usage

These files are intended for:

• Zero-shot EC function prediction (top1 and max-separation)
• Enzyme/non-enzyme binary classification based on RAMER embeddings
• Training/inference reproduction using the released scripts

Deployment / pipeline reference

For end-to-end scripts, deployment examples, and pipeline details, please refer to the GitHub organization:

• Ming-Ni-Group on GitHub

And the project repository:

• Ming-Ni-Group/RAMER

Notes

• This repository is primarily a resource host (weights + data + background library).
• Runtime scripts and workflow orchestration are maintained in the GitHub code repository.

License

The source code and model weights in this repository are licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0). See LICENSE for the full text.

Third-party base models (ProtT5, MolT5, GearNet) retain their original licenses. See THIRD_PARTY_MODELS.md for details.