Uncertainty-Aware Adaptation of Large Language Models for Protein-Protein Interaction Analysis

2025-02-10

AI Summary

To address high uncertainty and poor reproducibility of large language models (LLMs) in protein–protein interaction (PPI) prediction, this work introduces the first uncertainty quantification (UQ) adaptation framework integrating LoRA ensembling with Bayesian LoRA—enabling confidence-aware, reproducible predictions for biomedical LLM applications. Built upon fine-tuned LLaMA-3 and BioMedGPT, the framework jointly models parameter and predictive uncertainty, achieving state-of-the-art performance across multiple disease-related PPI identification tasks. Experiments demonstrate substantial improvements in prediction reliability, calibration accuracy, and biological interpretability. By unifying UQ with efficient parameter-efficient fine-tuning, our approach establishes a robust, verifiable paradigm for deploying LLMs in computational biology, directly supporting evidence-based, precision medicine decision-making.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Relational Probabilistic Models
• Multiagent Systems: Multiagent Systems under Uncertainty

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Identification of protein-protein interactions (PPIs) helps derive cellular mechanistic understanding, particularly in the context of complex conditions such as neurodegenerative disorders, metabolic syndromes, and cancer. Large Language Models (LLMs) have demonstrated remarkable potential in predicting protein structures and interactions via automated mining of vast biomedical literature; yet their inherent uncertainty remains a key challenge for deriving reproducible findings, critical for biomedical applications. In this study, we present an uncertainty-aware adaptation of LLMs for PPI analysis, leveraging fine-tuned LLaMA-3 and BioMedGPT models. To enhance prediction reliability, we integrate LoRA ensembles and Bayesian LoRA models for uncertainty quantification (UQ), ensuring confidence-calibrated insights into protein behavior. Our approach achieves competitive performance in PPI identification across diverse disease contexts while addressing model uncertainty, thereby enhancing trustworthiness and reproducibility in computational biology. These findings underscore the potential of uncertainty-aware LLM adaptation for advancing precision medicine and biomedical research.

Problem

• Adapt LLMs for protein-protein interaction analysis
• Address uncertainty in LLM predictions for reliability
• Enhance reproducibility in computational biology applications

Innovation

• Uncertainty-aware adaptation of LLMs
• Integration of LoRA ensembles
• Bayesian LoRA models for UQ