About Crux
A decision layer for target selection.
Crux turns public evidence into a single, cited brief on which biological targets are worth pursuing for a disease. It ends with a ranked recommendation, while showing the evidence, uncertainty and case against each target.
Why it exists
Why not just use Open Targets?
Open Targets is a data platform. It provides a broad, neutral view of target-disease evidence and the underlying signals that contribute to an association score.
Crux does not replace that platform or rebuild its upstream sources. Open Targets is one of Crux's core suppliers.
Crux competes with the manual analyst workflow: Open Targets in several tabs, ClinicalTrials.gov in another, literature searches elsewhere, and a spreadsheet used to assemble the conclusion. Crux turns that workflow into one document and ends with a defensible judgement.
The decision layer
Three things a neutral data platform does not do
01
Synthesises into a judgement
Open Targets exposes the evidence. Crux combines association strength, tractability, trials and literature into a written assessment of what the evidence means.
02
Adds the competitive landscape
Target biology is only part of the decision. Crux connects known drugs to ClinicalTrials.gov so the reader can see whether a field is open, active or already crowded.
03
States the case against and recommends
Every target includes a substantive argument against pursuing it. Crux then ranks the strongest opportunities based on the evidence and competitive context, rather than simply repeating the association score.
How it works
Public evidence, assembled before synthesis
Crux gathers evidence through a fixed TypeScript workflow. The language model does not decide which sources to search or introduce outside knowledge. It receives the assembled evidence, synthesises it into the brief and must cite every substantive claim.
When the available evidence is thin, the brief reports it as thin. Missing trials, weak tractability or literature-led associations are not filled in with assumptions.
- Open Targets
- Target-disease associations, evidence breakdowns, tractability and known drugs.
- ClinicalTrials.gov
- Linked clinical trials, phases, sponsors, interventions and status.
- Europe PMC
- Relevant scientific literature and abstracts used to ground the literature assessment.
Status
A working portfolio prototype
Crux was built by Daniel Johnson to demonstrate how public biomedical data, deterministic orchestration and constrained language-model synthesis can produce a credible decision-support tool. It is not a commercial product and should not replace programme-specific scientific judgement.