Zitnik Lab
Zitnik Lab is an academic research lab in Harvard Medical School's Department of Biomedical Informatics. It develops multimodal, generative and agentic AI that reasons across molecular, cellular, tissue and patient scales to advance scientific discovery, therapeutic design, individualized diagnosis and treatment.
At a Glance
- Universities and academic research laboratories
- Biotechnology and pharmaceutical R&D
- Clinicians, hospitals and clinical-research organizations
- Computational biology, genomics, proteomics and drug-discovery researchers
- +1 more
AI Tools by Zitnik Lab
(1)
ToolUniverse
Scientific Tools for LLM Agents
Discussions
No discussions yet
Be the first to start a discussion about Zitnik Lab
Latest News
Five Zitnik Lab papers accepted to NeurIPS 2026, including AutoScientists, biological reasoning models, STRAND and StereoPep.
World Models for Biomedicine published/announced, describing AI systems that simulate biological systems under interventions.
Launch and research announcement for Open AI Scientists, an initiative to develop autonomous AI systems that reason, hypothesize and experiment alongside researchers.
COMPASS pan-cancer foundation model reported for predicting immunotherapy response.
Products & Services
Open academic ecosystem and Python/MCP platform that lets any LLM operate as an AI scientist through standardized tool discovery, calling, composition and optimization; it integrates 1,000+ scientific resources (the documentation also describes 2,700+ tools in the CLI). Available through PyPI, GitHub, a CLI, Python API and agent integrations.
AI agent for therapeutic reasoning across a universe of biomedical tools, powered by ToolUniverse.
Omics AI agent for therapeutic discovery that nominates therapeutic targets across cell contexts, predicts synthetic lethality and forecasts immunotherapy response from multimodal patient profiles.
AI agent for treatment reasoning over all FDA-approved drugs since 1939, trained with reinforcement learning over a biomedical tool universe.
Market Position
Zitnik Lab is positioned as an academic AI-for-science and AI-for-medicine research group rather than a conventional software company. Its distinctive focus is on scientifically grounded multimodal, geometric and agentic systems; ToolUniverse differentiates from general-purpose agent frameworks and individual scientific APIs by unifying 1,000+ live scientific tools/databases with standardized discovery, calling and composition for AI scientists.
Leadership
Founders
Marinka Zitnik
Associate Professor of Biomedical Informatics at Harvard Medical School; Associate Faculty at Harvard's Kempner Institute, Associate Member of the Broad Institute, and Faculty Affiliate of Harvard Data Science. Before Harvard she was a Stanford Computer Science postdoctoral scholar and Chan Zuckerberg Biohub researcher; she earned a PhD in Computer Science from the University of Ljubljana and researched at Imperial College London, University of Toronto and Baylor College of Medicine.
Executive Team
Marinka Zitnik
Lab Director; Associate Professor of Biomedical Informatics, Harvard Medical School
Stanford Computer Science postdoctoral scholar and Chan Zuckerberg Biohub researcher; PhD in Computer Science from the University of Ljubljana; Associate Faculty at the Kempner Institute and Associate Member of the Broad Institute.
Founding Story
Marinka Zitnik announced that she would join Harvard as a tenure-track Assistant Professor in December 2019 and establish a laboratory focused on Artificial Intelligence for Science, Medicine and Health. The initial vision was to develop machine-learning methods for rich biomedical interaction data and translate them into genomics, drug-discovery, health and medical applications.
Business Model
Revenue Model
Academic research lab whose projects are primarily research and open-source initiatives. ToolUniverse is described as an open science, academic research project and is distributed through public GitHub, PyPI, documentation and agent integrations rather than a published commercial subscription model.
Target Markets
- Universities and academic research laboratories
- Biotechnology and pharmaceutical R&D
- Clinicians, hospitals and clinical-research organizations
- Computational biology, genomics, proteomics and drug-discovery researchers
- Developers building AI agents with scientific tools
- Drug discovery, drug design and drug repurposing
- Cancer biomarker discovery, target identification, drug-combination selection and immunotherapy response prediction
- Precision medicine, individualized diagnosis and treatment reasoning
- Multi-omics, single-cell and spatial biology analysis
- Biomedical literature and knowledge-graph exploration
- Autonomous scientific hypothesis generation and experiment planning
- Harvard Medical School and Harvard University research community
- Broad Institute of MIT and Harvard research community
- Roche computational and biomarker scientists
- Researchers and developers using the open ToolUniverse ecosystem