From Single to Multi-Agent Reasoning: Advancing GeneGPT for Genomics QA

TL;DR

This paper introduces GenomAgent, a flexible multi-agent framework that improves genomic question answering by coordinating specialized agents, outperforming the state-of-the-art GeneGPT and adaptable to other scientific domains.

Contribution

We develop GenomAgent, a novel multi-agent system that overcomes GeneGPT's API rigidity, enhancing complex genomics QA and demonstrating broader applicability across scientific fields.

Findings

GenomAgent outperforms GeneGPT by 12% on average across nine tasks.

The framework demonstrates adaptability beyond genomics to other scientific domains.

GenomAgent effectively coordinates specialized agents for complex queries.

Abstract

Comprehending genomic information is essential for biomedical research, yet extracting data from complex distributed databases remains challenging. Large language models (LLMs) offer potential for genomic Question Answering (QA) but face limitations due to restricted access to domain-specific databases. GeneGPT is the current state-of-the-art system that enhances LLMs by utilizing specialized API calls, though it is constrained by rigid API dependencies and limited adaptability. We replicate GeneGPT and propose GenomAgent, a multi-agent framework that efficiently coordinates specialized agents for complex genomics queries. Evaluated on nine tasks from the GeneTuring benchmark, GenomAgent outperforms GeneGPT by 12% on average, and its flexible architecture extends beyond genomics to various scientific domains needing expert knowledge extraction.