AI evidence extraction

The next paradigm in bioinformatics: a review of multi-agent systems and foundational models for end-to-end scientific discovery

Francesco Branda, Mohamed Mustaf Ahmed, Massimo Ciccozzi, Pietro Hiram Guzzi, Fabio Scarpa · Briefings in Bioinformatics · 2026

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

9/10

Relevance

0/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1093/bib/bbag245

Methodology & findings

Study design

Narrative literature review synthesizing recent developments in foundational models, multi-agent systems, and their applications in bioinformatics and drug discovery.

Main result

The review argues that "the next paradigm shift will go beyond traditional predictive models and generative artificial intelligence (AI) toward agentic AI: systems capable of planning, acting through tools, reflecting on results, and iterating until a goal is achieved." The paper identifies foundational models such as scGPT, Nicheformer, and EpiAgent as key developments enabling transferable representations across omic modalities, while biomedical agent frameworks like ClinicalAgent and Biomni operationalize agentic principles in controlled environments.

Research paradigm

Systems-based integrative approach combining computational methods, artificial intelligence, and bioinformatics theory

Author conclusions

The authors conclude that "the main challenges ahead" include "hallucinations, interpretability, systemic biases, integration with clinical infrastructures, and regulatory and ethical requirements," and they "propose a roadmap for the development of scientific agents that are not only high-performing but also reliable, verifiable, and implementable in real biomedical contexts."

Limitations

The authors identify that "the main challenges ahead" include "hallucinations, interpretability, systemic biases, integration with clinical infrastructures, and regulatory and ethical requirements." The review notes these as barriers to developing scientific agents that are "reliable, verifiable, and implementable in real biomedical contexts."

Open questions raised

The review identifies critical gaps including: hallucinations in AI-generated outputs, interpretability constraints in foundational models, systemic biases in training data and agent decision-making, integration challenges with existing clinical infrastructures, regulatory compliance requirements, and ethical considerations for agentic systems in healthcare settings
Integration of universal biological models with multi-agent systems for end-to-end scientific discovery; development of agentic AI systems with improved hallucination mitigation; enhancement of interpretability in foundational models; addressing systemic biases; integration with clinical infrastructures; and meeting regulatory and ethical requirements for real-world biomedical implementation.
The review identifies key future research directions including: addressing hallucinations in agentic systems, improving interpretability of foundational models, mitigating systemic biases, achieving seamless integration with clinical infrastructures, and satisfying regulatory and ethical requirements for deployment in real biomedical contexts.

Data: not_statedCode: not_statedExtracted from: pdfAgreement 63%

Explore related topics

Epistemic Risk Research Integrity Literature Review