HypoEvolve: Genetic Algorithms Enable Multi-Agent LLMs to Discover Scientific Hypotheses

arXiv 2026 52.1 method

TLDR

Introduces HypoEvolve, a genetic algorithm coordinating multi-agent LLMs to discover and refine scientific hypotheses, with evaluations on drug repurposing using external evidence.

Reasoning

The paper presents a clear framework for studying how agent collaboration affects hypothesis quality, with a strong design for separating agent capabilities from collaboration effects. Its main limitation is the narrow application to drug repurposing, and the abstract does not report detailed experimental results or baselines.

Read-first score

Read-first score 52.1, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 54.

Recency 8%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=baseline,evaluation,experiment

Topical relevance 42%

Uses existing LLM keyword relevance scores normalized to 0-100. AI scientist,automated scientific discovery,autonomous research agent,automated research,literature review agent,survey generation,automated experimentation,experiment design agent,AI for scientific research,paper writing agent,research automation,scientific discovery agent

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Field roles

FrontierMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 43.

Keyword Scores

automated scientific discovery

scientific discovery agent

AI for scientific research

AI scientist

automated research

research automation

autonomous research agent

experiment design agent

literature review agent

automated experimentation

survey generation

paper writing agent

Deep Analysis

Innovations

HypoEvolve makes agent collaboration explicit through successive updates to a hypothesis population.
A generational genetic algorithm coordinates specialized LLM agents that integrate mechanistic arguments, reconsider assumptions, and assess evidence and testability.
The framework directly tests how collaboration affects hypothesis quality by specifying how scientific judgments and new proposals reshape the population each generation.
Evaluation uses scientifically meaningful drug repurposing hypotheses linked to target-level biological claims.
DepMap and Open Targets are adapted into complementary external measures grounded in experimental, genetic, and clinical evidence.

Methodology

HypoEvolve uses a generational genetic algorithm to coordinate specialized LLM agents that integrate mechanistic arguments, reconsider assumptions, and assess evidence and testability, with each generation updating a hypothesis population. It is evaluated on drug repurposing hypotheses across 34 cancer types using adapted DepMap and Open Targets measures, compared against six baselines.

Key Results

Across 34 cancer types, HypoEvolve achieves the highest scores against six baselines on both external measures, with DepMap selectivity reaching 0.171 versus 0.115 for the strongest baseline. Gains over single-pass generation also generalize to held-out cancer types.

Resources

Paper