Large Language Models for Non-Coding RNA Biomarker Discovery in Breast Cancer

Breast cancer (BC) remains a leading cause of cancer-related mortality globally, driven by its molecular heterogeneity and complex tumor biology. Early and accurate detection is critical for improving patient survival and guiding targeted therapies. Noncoding RNAs (ncRNAs), particularly circular RNAs (circRNAs), small nucleolar RNAs (snoRNAs), and PIWI-interacting RNAs (piRNAs), have emerged as key regulators in cancer progression and as potential biomarkers. However, the sheer volume of omics data and literature poses challenges in extracting actionable insights. Recent advances in large language models (LLMs) offer new opportunities to accelerate biomarker discovery through semantic reasoning, knowledge integration, and pattern recognition. This chapter explores the integration of LLMs with ncRNA biology, focusing on circRNAs, snoRNAs, and piRNAs in BC, and proposes a computational framework for biomarker identification.