Back

GeneRAG: Enhancing Large Language Models with Gene-Related Task by Retrieval-Augmented Generation

Lin, X.; Deng, G.; Li, Y.; Ge, J.; Ho, J. W. K.; Liu, Y.

2024-06-28 bioinformatics
10.1101/2024.06.24.600176 bioRxiv
Show abstract

Large Language Models (LLMs) like GPT-4 have revolutionized natural language processing and are used in gene analysis, but their gene knowledge is incomplete. Fine-tuning LLMs with external data is costly and resource-intensive. Retrieval-Augmented Generation (RAG) integrates relevant external information dynamically. We introduce GO_SCPLOWENEC_SCPLOWRAG, a frame-work that enhances LLMs gene-related capabilities using RAG and the Maximal Marginal Relevance (MMR) algorithm. Evaluations with datasets from the National Center for Biotechnology Information (NCBI) show that GO_SCPLOWENEC_SCPLOWRAG outperforms GPT-3.5 and GPT-4, with a 39% improvement in answering gene questions, a 43% performance increase in cell type annotation, and a 0.25 decrease in error rates for gene interaction prediction. These results highlight GO_SCPLOWENEC_SCPLOWRAGs potential to bridge a critical gap in LLM capabilities for more effective applications in genetics.

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.