Positioning Multi-Agent Large Language Models as the Future of Bulk Gene Expression Analysis and Cancer Prediction

SeyedMohsen Hosseini, Divya Sharma

Keywords: LLM, multimodal, bioinformatic, agentic ai

Abstract: Bulk RNA sequencing is essential for understanding cancer biology, yet current computational methods struggle with cross-cohort generalization, interpretability, and multi-source integration. We propose multi-agent architectures built from specialized large language models (LLMs) as a solution. Unlike monolithic models, our framework integrates multi-modal inputs. By assigning complementary tasks to expression-focused, sequence-based, literature-aware, and integrative agents, the system achieves more robust, interpretable, and clinically meaningful insights. We discuss supporting evidence, potential challenges, and a research agenda, emphasizing the paradigm’s importance for precision oncology.

Submission Number: 67