BioDyad: Synchronize Biomedical Discovery and Machine Learning Engineering

Xingbo Du; Fadli Aulawi Al Ghiffari; Leonard Song; Loka Li; Duzhen Zhang; Zixiao Wang; Xiuying Chen; Le Song

cs.AISep 25, 2026

BioDyad: Synchronize Biomedical Discovery and Machine Learning Engineering

Authors: Xingbo Du, Fadli Aulawi Al Ghiffari, Leonard Song, Loka Li, Duzhen Zhang, Zixiao Wang, Xiuying Chen, Le Song

Organizations: Mohamed bin Zayed University of Artificial Intelligence, Abu Dhabi, UAE · The Westminster Schools, Atlanta, GA, USA

Abstract

Agentic biomedical machine learning (ML) draws on complementary advances in biomedical evidence acquisition and executable program search. Existing systems connect aspects of these capabilities, but coordinating them throughout program search remains challenging. New evidence must guide candidate construction, execution outcomes must inform subsequent discovery and reuse, and validation demands must fit the search budget. We introduce BioDyad, which couples biomedical discovery and ML engineering through two hierarchies within Monte Carlo graph search. Its scientific hierarchy combines prior biomedical guidance with iterative discovery, then links biomedical plans to execution outcomes in memory for reuse across candidates. Its engineering hierarchy moves candidate programs from smoke execution, through train/validation evaluation, to full-data retraining. We evaluate BioDyad on the 76-task BioXArena benchmark under a two-hour per-task budget with three matched LLM backends. It achieves the highest penalized all-task score and task success rate among four agent methods and a one-shot baseline under each backend. These results support coordinating biomedical discovery and ML engineering to integrate external knowledge into executable programs across heterogeneous biomedical tasks.

Figures & tables

Type	Method	Scientific hierarchy		Search and engineering
Type	Method	Bio. discovery	Bio. memory	Program search	Staged eval.	Budget adapt.
Harness	BioXArena harness ( Li et al., 2026 )	✗	✗	✗	✗	✗
Bio. agents	Biomni ( Huang et al., 2025 )	✓	✗	✗	✗	✗
	STELLA ( Jin et al., 2025 )	✓	✗	✗	✗	✗
	VCHarness ∗ ( Cheng et al., 2026 )	$\triangle$	$\triangle$	✓	✗	✗
MLE agents	MLEvolve ( Du et al., 2026 )	✗	✗	✓	✗	✓

Table 1: Capabilities of representative agentic systems for biomedical ML program discovery.

Figure 1: BioDyad couples biomedical discovery and ML engineering within MCGS. The scientific hierarchy combines prior guidance, node-level discovery, and biomedical memory; the engineering hierarchy advances programs through smoke execution, train/validation evaluation, and full-data runs. Biomedical evidence guides candidate expansion, while execution feedback updates the search graph and memory within the wall-clock budget. The hERG Binding Affinity task from BioXArena ( Li et al., 2026 ) provides an illustrative example.

Figure 2: ReAct-based prior biomedical guidance. A plan-only agent alternates reasoning with read-only probes to compile verified data findings and biomedical insights.

Figure 3: Success rate (x-axis) and penalized all-task score (y-axis) under three LLM backends. Each panel compares four agent methods with a two-hour per-task budget and a one-shot baseline retaining only the first attempt, evaluated on the same 76 tasks.

LLM backend	Method	Seq	SC	Str	Net	Chem	Pert	Phen	Img	Text	Avg.
GPT-5.6-sol	One-shot	0.416	0.288	0.727	0.458	0.576	0.291	0.508	0.500	0.546	0.472
	BioXArena Harness	0.663	0.786	0.834	0.458	0.928	0.517	0.606	0.714	0.637	0.685
	Biomni	0.773	0.652	0.591	0.490	0.690	0.404	0.431	0.702	0.656	0.605
	MLEvolve	0.790	0.798	0.794	0.527	0.922	0.597	0.575	0.673	0.720	0.715
	BioDyad (ours)	0.776	0.861	0.815	0.521	0.926	0.776	0.594	0.666	0.671	0.738
GLM-5.1	One-shot	0.187	0.102	0.236	0.062	0.556	0.121	0.374	0.139	0.030	0.198

Table 2: Domain breakdowns of four agent methods and a one-shot baseline under three matched LLM backends on the 76 BioXArena tasks. Scores are penalized average scores; higher is better. Avg. is weighted by task count across all 76 tasks. Bold and underlining indicate the highest and second-highest distinct displayed scores, respectively, within each backend and column.

Method	Penalized score	Success rate
BioXArena Harn.	$0.699\,{\scriptstyle\pm 0.005}$	$\mathbf{0.944}\,{\scriptstyle\pm 0.000}$
Biomni	$0.647\,{\scriptstyle\pm 0.012}$	$0.889\,{\scriptstyle\pm 0.000}$
MLEvolve	$0.688\,{\scriptstyle\pm 0.019}$	$0.907\,{\scriptstyle\pm 0.032}$
BioDyad	$\mathbf{0.711}\,{\scriptstyle\pm 0.002}$	$\mathbf{0.944}\,{\scriptstyle\pm 0.000}$

Table 3: Repeated-run robustness on the fixed 18-task subset with GPT-5.6-sol. Values are mean

\pm

sample standard deviation (SD) across three independent runs.

Hierarchy	Mechanism	MLEvolve	MLEvolve	BioDyad	BioDyad	BioDyad	BioDyad
Hierarchy	Mechanism	MLEvolve	w/ discovery	w/o memory	w/o Sci.	w/o Eng.	(full)
Sci. hierarchy	Bio. discovery	✗	✓	✓	✗	✓	✓
Sci. hierarchy	Bio. memory	✗	✗	✗	✗	✓	✓
Eng. hierarchy	Staged evaluation	✗	✗	✓	✓	✗	✓
Performance $\uparrow$	Chemical biology	0.898	0.795	0.913	0.861	0.896	0.919
	Phenotype-disease	0.538	0.599	0.560	0.468	0.570	0.606

Table 4: Component ablations with the GPT-5.4 backend. Scores are penalized task averages. Avg. is the task-weighted mean across all 16 tasks in the two domains; higher is better.

Figure 4: Biomedical feature integration in BioDyad: (a) predictive gain, (b) smoke-guided execution repair, and (c) no predictive gain despite successful execution. Scores are task-specific validation metrics; full execution denotes candidate train/validation evaluation.

Domain	Best-so-far test score						Score gain
Domain	2 h	4 h	6 h	8 h	10 h	12 h	2-4 h	4-12 h	2-12 h
Chemical biology	0.927	0.932	0.936	0.936	0.936	0.937	+0.006	+0.004	+0.010
Phenotype-disease	0.640	0.663	0.688	0.690	0.690	0.690	+0.022	+0.027	+0.050

Table 5: Retrospective best-so-far normalized test scores during 12-hour BioDyad runs with GPT-5.6-sol. Each checkpoint averages task-level maxima over all eight tasks in a domain, using the same runs as Appdx. D . Gain columns report absolute differences computed before rounding.

Figure 5: Candidate test scores during 12-hour searches: (a) BACE1 binding affinity, (b) CYP inhibition, (c) Alzheimer’s disease staging, (d) breast-cancer subtype. Gray points show candidate scores; blue lines show retrospective maxima; orange dashed lines show linear fits against time.

Appendix figures & tables7 assets

Supplementary material from the paper’s appendix.

Appendix

Agent	Seq	SC	Str	Net	Chem	Pert	Phen	Img	Text	Avg.
BioDyad (GPT-5.6-sol)	0.776	0.861	0.815	0.521	0.926	0.776	0.594	0.666	0.671	0.738
BioDyad (GLM-5.1)	0.776	0.771	0.729	0.509	0.932	0.737	0.600	0.700	0.552	0.705
BioDyad (DeepSeek-V4)	0.717	0.791	0.789	0.510	0.909	0.717	0.600	0.646	0.608	0.701
MLEvolve ( Du et al., 2026 )	0.765	0.798	0.750	0.510	0.869	0.601	0.502	0.606	0.534	0.666
STELLA ( Jin et al., 2025 )	0.677	0.559	0.481	0.497	0.878	0.555	0.574	0.651	0.641	0.613
Biomni ( Huang et al., 2025 )	0.526	0.646	0.569	0.497	0.723	0.628	0.547	0.659	0.415	0.579

Appendix

Table 6: Comparison with the penalized score presented on BioXArena’s leaderboard across all 76 tasks. Avg. is task-weighted across all tasks. Baseline scores are reported by BioXArena ( Li et al., 2026 ) ; BioDyad scores use the same evaluation protocol. Bold marks the highest score in each column. Backends differ across systems (Appdx. E.2 ).

Hierarchy	Mechanism	MLEvolve	MLEvolve	BioDyad	BioDyad	BioDyad	BioDyad
Hierarchy	Mechanism	MLEvolve	w/ discovery	w/o memory	w/o Sci.	w/o Eng.	(full)
Sci. hierarchy	Bio. discovery	✗	✓	✓	✗	✓	✓
Sci. hierarchy	Bio. memory	✗	✗	✗	✗	✓	✓
Eng. hierarchy	Staged evaluation	✗	✗	✓	✓	✗	✓
Outcomes	Full runs	552	672	639	446	668	566
	Valid	304 (55.1%)	314 (46.7%)	425 (66.5%)	319 (71.5%)	411 (61.5%)	458 (80.9%)

Appendix

Table 7: Candidate outcomes in the component ablations. Each entry gives a count and its percentage of that configuration’s full-execution runs.

Domain	Task	Test 2 h	Test 4 h	Test 12 h	$\Delta$ 2-4 h	$\Delta$ 4-12 h
CB	BACE1 Binding Affinity	0.934	0.936	0.939	+0.002	+0.003
CB	Cell Painting Perturbation	0.982	1.000	1.000	+0.018	+0.000
CB	CYP Inhibition (Multi-label)	0.940	0.941	0.944	+0.001	+0.003
CB	EGFR Binding Affinity	0.925	0.927	0.927	+0.003	+0.000
CB	GPCR Binding (Multi-class)	0.948	0.953	0.970	+0.005	+0.017
CB	hERG Binding Affinity	0.898	0.908	0.909	+0.010	+0.001

Appendix

Table 8: Normalized test-score checkpoints and gains in the 12-hour traces. CB denotes chemical biology and PD denotes phenotype-disease. Each checkpoint is a retrospective maximum over completed candidates. Gains are absolute differences computed before rounding.

Figure 6: Cumulative best normalized test scores for all 16 tasks, using the same runs as Table 8 . The upper two rows show chemical-biology tasks and the lower two show phenotype-disease tasks. Each task has its own vertical scale. Steps occur at logged candidate completion; dashed guides mark 2 and 4 hours. No score is shown before the first completed candidate, and the last observed best is carried forward to 12 hours.

Domain	First task	Second task
Sequence	gene-tissue-expression	isoform-expression
Single-cell	batch-integration	cell-type-from-expression
Structure	complex-structure-evaluation	enzyme-commission-prediction
Network biology	gene-disease-association	go-function-multi-label
Chemical biology	bace1-binding-affinity	cell-painting-perturbation
Perturbation dynamics	cancer-drug-sensitivity	crispr-perturbation-prediction

Appendix

Table 9: Fixed repeated-run subset: the first two tasks per domain in BioXArena Table A4, preserving its within-domain order.

Method	Metric	Run 1	Run 2	Run 3	Mean $\pm$ SD
BioXArena Harness	Score	0.695	0.705	0.699	$0.699{\scriptstyle\pm 0.005}$
BioXArena Harness	Success rate	0.944	0.944	0.944	$0.944\,{\scriptstyle\pm 0.000}$
Biomni	Score	0.637	0.660	0.644	$0.647{\scriptstyle\pm 0.012}$
Biomni	Success rate	0.889	0.889	0.889	$0.889{\scriptstyle\pm 0.000}$
MLEvolve	Score	0.709	0.683	0.672	$0.688{\scriptstyle\pm 0.019}$
MLEvolve	Success rate	0.944	0.889	0.889	$0.907{\scriptstyle\pm 0.032}$

Appendix

Table 10: Per-run results on the fixed 18-task subset with GPT-5.6-sol. Score is the penalized mean over all 18 tasks; success rate is the successful task count divided by 18. Summary entries report mean

\pm

sample standard deviation (SD) across three independent runs.

Method	LLM backend
Biomni	Claude Sonnet 4 ( Anthropic, 2025 )
STELLA	Dev/Tool-Creation: Claude Sonnet 4.6 ( Anthropic, 2026b ) ;
	Manager/Critic: Gemini 3.1 Pro ( The Gemini Team, 2026 )
MLEvolve	Gemini 3.1 Pro ( The Gemini Team, 2026 )
ML-Master 2.0	DeepSeek-V4-Pro ( Xu et al., 2026 )

Appendix

Table 11: Original backends of BioXArena’s reported agent systems, retained for the overall reference experiment.

AI Agents for Scientific Discovery

AI agents for scientific discovery use language models, tools, and experiments to generate, test, and refine scientific knowledge.

See all 288 papers

BioResearcher: Scenario-Guided Multi-Agent for Translational Medicine
May 7, 2026Remigiusz Kinas, Joanna Krawczyk, Rafał Powalski +6Multi-Agent Orchestration AI Agents for Scientific Discovery
Ingenix.AI, Warsaw, Poland
Beyond Prompt-Based Planning: MCP-Native Graph Planning-based Biomedical Agent System
Jun 3, 2026Zhangtianyi Chen, Florensia Widjaja, Wufei Dai +3Model Context Protocol Graph-Based Planning
1The Chinese University of Hong Kong, Shenzhen
BioMedArena: An Open-source Toolkit for Building and Evaluating Biomedical Deep Research Agents
May 7, 2026Jinge Wu, Hongjian Zhou, Mingde Zeng +8AI Agents for Scientific Discovery Deep Research Agents
University of Oxford · University College London · Technical University of Munich +1

May 7, 2026J/K move · Enter open · S save

BioResearcher: Scenario-Guided Multi-Agent for Translational Medicine

Remigiusz Kinas, Joanna Krawczyk, Rafał Powalski, Przemysław Pietrzak+5

Ingenix.AI, Warsaw, Poland

Multi-Agent Orchestration AI Agents for Scientific Discovery Healthcare Scientific Workflow Automation

Open paper AskPDF arXiv

Translational medicine turns underspecified development goals into evidence synthesis that must combine literature, trials, patents, and quantitative multi-omics analysis while preserving identifiers, uncertainty, and retrievable provenance. General-purpose foundation models and off-the-shelf tool-augmented or multi-agent systems are not built for this: they tend to produce single-shot answers or run open-endedly, and fall short on the auditable, scenario-specific workflows that heterogeneous biomedical sources demand. This paper introduces Ingenix BioResearcher, a scenario-guided multi-agent system that maps queries to versioned research playbooks, delegates to specialized subagents over 30+ tools and machine-learning endpoints, mixes structured database access with sandboxed code for genome-scale analyses, and applies claim-level multi-model reconciliation before editorial assembly. We evaluate BioResearcher across unit-level capabilities, open-ended biomedical reasoning, and end-to-end clinical discovery. It leads evaluated baselines on 109 single-step tests (83.49% pass rate; 0.892 average score), achieves strong biomedical benchmark performance (89.33% on BixBench-Verified-50 and the top 0.758 mean score on BaisBench Scientific Discovery), and leads on a 30-query clinical end-to-end benchmark with the highest positive hit rate (74.7%

\pm

3.3%) and negative clear rate (96.8%

\pm

0.2%). These results show broad, competitive performance across unit-level, open-ended, and end-to-end clinical evaluations.

Abstract

AI Agents for Scientific Discovery

Explore similar work

BioResearcher: Scenario-Guided Multi-Agent for Translational Medicine

Beyond Prompt-Based Planning: MCP-Native Graph Planning-based Biomedical Agent System

BioMedArena: An Open-source Toolkit for Building and Evaluating Biomedical Deep Research Agents