About Robertium
Why this project exists and where it's going
The problem
Drug discovery is slow and expensive. The average new drug takes 10–15 years and over $2 billion to bring to market. Yet many of the connections that could lead to new treatments already exist in the published literature — they just remain hidden in the noise of 4,000+ biomedical papers published every day.
Drug repurposing — finding new uses for existing approved drugs — is a faster, cheaper alternative. But systematic discovery of repurposing candidates requires reading and connecting evidence across thousands of papers in different therapeutic domains. No human team can do this at scale.
Robertium is built to do exactly that.
Who built this
Robertium is built and maintained by Daniel Trofimov as an independent open-source project, with active commits to github.com/routewise96/robertium and a public methodology log in /updates.
My background is in engineering, not biomedicine. I came to this problem from the infrastructure side: structured literature mining at scale is a tractable software problem, and the existing open tools stop short of cross-domain hypothesis generation. Every methodological decision is made in public, with input from researchers who know more than I do about the biology, and corrections from the community land as commits, not silent edits.
Daniel is on GitHub and LinkedIn. The project name honors my father, who taught me that meaningful work takes years and that patience is a discipline.
The approach
Robertium reads open biomedical literature, extracts structured claims, and applies a 40-year-old methodology — Don R. Swanson's literature-based discovery — using modern language models and knowledge graphs.
The result: candidate hypotheses with full evidence chains, ready for expert review and experimental validation. Not predictions, not magic — just systematic discovery of connections that already exist in published research.
Read the methodology in detail at /method.
Why open-source
Drug discovery infrastructure is too important to be locked behind proprietary platforms. We believe:
- The biomedical literature is a public good. Tools to navigate it should be public too.
- Reproducibility is fundamental to science. Closed systems break that.
- Researchers should be able to inspect, modify, and improve the methods that surface hypotheses they will spend years validating.
All code is MIT licensed. All extracted data will be released under CC-BY-4.0 alongside the first preprint. The knowledge graph is downloadable. Anyone can clone the repo and run the full pipeline on their own domain.
What's next
2026 (so far)
- Multi-domain corpus live: 10 therapeutic domains, 167,145 papers, 270,791 structured claims
- Cross-domain hypothesis catalog: 24,285 candidates total, 1,618 passing outreach-quality filters after dedup (197 high-confidence) with full evidence chains
2026 (in progress)
- Public preprint on bioRxiv describing methodology and first findings
- Open dataset release for the research community under CC-BY-4.0
- Outreach to domain experts for review of top candidates
2027
- Expanded corpus: 100,000+ papers per major domain
- Active collaboration with experimental labs for validation of top candidates
- Public-facing tools: API for researchers, web interface for hypothesis exploration
- Funding through open science grants (Mozilla, Wellcome, CZI)
Beyond
- Build a sustainable infrastructure that biomedical researchers rely on
- Integrate with clinical data, drug-target databases (DrugBank, ChEMBL)
- Train domain-specific models for higher extraction precision
- Establish Robertium as standard infrastructure for literature-based discovery
Get involved
Robertium grows through collaboration. I'm looking for:
- Domain experts — Oncologists, neurologists, pharmacologists who can review top hypotheses in their area and tell us what's interesting vs trivial.
- Computational biologists — Especially those working with knowledge graphs, NLP for biomedicine, or LBD methodology.
- Engineers — Open-source contributions, performance improvements, new features.
- Funders — If your foundation supports open biomedical infrastructure.
Reach out at daniel@robertium.com or via GitHub.