MULTI-AGENT AI IN GENETICS AND MOLECULAR BIOLOGY: A REVIEW OF EMERGING ARCHITECTURES, APPLICATIONS AND FUTURE DIRECTIONS

Abstract

Advancements in Artificial Intelligence (AI) particularly Large Language Models (LLMs), have accelerated the development of Multi-Agent Artificial Intelligence (MAAI) systems in biomedical research. In contrast to the traditional single-agent systems multi-agent system consists of multiple specialized agents which perform reasoning, tool usage, data retrieval, hypothesis generation, workflow orchestration and validation in a collaborative manner. This review analyses developments of multi-agent AI use in genetics and molecular biology with focus on genomics, transcriptomics, bioinformatics, drug discovery, plant biology and biomedical scientific discovery. Different important frameworks like Robin, OpenBioLLM, GenomAgent, Coated-LLM, PhenoAssistant and oncology-focused genomic systems are analyzed to understand their architectures, functionalities, advantages, and limitations. Existing studies demonstrate how multi-agent systems helps to improve genomic reasoning, to automate bioinformatics workflows, to enhance biological data interpretation and also helps for scientific hypothesis generation. The review also highlights the growing integration of LLMs with retrieval systems, biological databases and multi-omics platforms. Though the results are encouraging, still there are challenges which includes limited biological validation, lack of standardized benchmarks, explainability concerns, reproducibility issues, computational cost and insufficient real-world deployment. Evidence advocates that multi-agent AI has potential to transform genetics and molecular biology by enabling autonomous scientific assistants, adaptive reasoning systems, and intelligent bioinformatics platforms. Future research should move on the direction of on trustworthy AI, clinically validated systems, multi-omics reasoning, standardized evaluation protocols and safer autonomous scientific workflows.