Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

VenusFactory: A Unified Platform for Protein Engineering Data Retrieval and Language Model Fine-Tuning

Yang Tan, Chen Liu, Jingyuan Gao, Banghao Wu, Mingchen Li, Ruilin Wang, Lingrong Zhang, Huiqun Yu, Guisheng Fan, Liang Hong, Bingxin Zhou

TL;DR

The paper tackles the fragmentation of AI-driven protein engineering workflows by uniting data retrieval, benchmarking, and model fine-tuning in a single platform. It introduces VenusFactory, an extensible engine that integrates rapid data collection from major databanks, a standardized benchmarking suite across five protein-engineering tasks, and modular fine-tuning pipelines for a wide range of PLMs. The system supports both CLI and Gradio interfaces and hosts 40+ datasets and 40+ PLMs, enabling researchers from biology and computer science to prototype end-to-end solutions. Empirical results show SES-Adapter often achieves top performance across tasks, demonstrating the value of task-specific, structure-aware fine-tuning for protein engineering. This platform promises to accelerate interdisciplinary AI-driven protein engineering by simplifying data access, benchmarking, and model adaptation.

Abstract

Natural language processing (NLP) has significantly influenced scientific domains beyond human language, including protein engineering, where pre-trained protein language models (PLMs) have demonstrated remarkable success. However, interdisciplinary adoption remains limited due to challenges in data collection, task benchmarking, and application. This work presents VenusFactory, a versatile engine that integrates biological data retrieval, standardized task benchmarking, and modular fine-tuning of PLMs. VenusFactory supports both computer science and biology communities with choices of both a command-line execution and a Gradio-based no-code interface, integrating $40+$ protein-related datasets and $40+$ popular PLMs. All implementations are open-sourced on https://github.com/tyang816/VenusFactory.

VenusFactory: A Unified Platform for Protein Engineering Data Retrieval and Language Model Fine-Tuning

TL;DR

The paper tackles the fragmentation of AI-driven protein engineering workflows by uniting data retrieval, benchmarking, and model fine-tuning in a single platform. It introduces VenusFactory, an extensible engine that integrates rapid data collection from major databanks, a standardized benchmarking suite across five protein-engineering tasks, and modular fine-tuning pipelines for a wide range of PLMs. The system supports both CLI and Gradio interfaces and hosts 40+ datasets and 40+ PLMs, enabling researchers from biology and computer science to prototype end-to-end solutions. Empirical results show SES-Adapter often achieves top performance across tasks, demonstrating the value of task-specific, structure-aware fine-tuning for protein engineering. This platform promises to accelerate interdisciplinary AI-driven protein engineering by simplifying data access, benchmarking, and model adaptation.

Abstract

Natural language processing (NLP) has significantly influenced scientific domains beyond human language, including protein engineering, where pre-trained protein language models (PLMs) have demonstrated remarkable success. However, interdisciplinary adoption remains limited due to challenges in data collection, task benchmarking, and application. This work presents VenusFactory, a versatile engine that integrates biological data retrieval, standardized task benchmarking, and modular fine-tuning of PLMs. VenusFactory supports both computer science and biology communities with choices of both a command-line execution and a Gradio-based no-code interface, integrating protein-related datasets and popular PLMs. All implementations are open-sourced on https://github.com/tyang816/VenusFactory.

Paper Structure

This paper contains 36 sections, 1 figure, 8 tables.

Table of Contents

  1. Introduction
  2. Data Collection
  3. Databanks
  4. Multithreaded Downloading
  5. Structure Serialization
  6. Task Benchmarking
  7. Localization
  8. Solubility
  9. Annotation
  10. Mutation
  11. Other Properties
  12. Model Application
  13. Pre-trained PLMs
  14. Collate Function
  15. Normalization
  16. ...and 21 more sections

Figures (1)

  • Figure 1: VenusFactory supports high-throughput raw data download, structure sequencing, a wide range of downstream task datasets, and interface or command-line protein language model fine-tuning and reasoning.