Direction, not just rank
Every therapy match says whether it is sensitivity, resistance, or another signal. A ranked list without direction is a trap.
Research-use-only variant intelligence
From variant to therapy, with the evidence in view.
Drugmarker turns a VCF into a reviewable case: annotated variants, cited therapy evidence, resistance signals, and a report your tumor board can actually use.
Pilot onboarding is reviewed manually. No patient data is required to evaluate.
EGFR
ClinVar: drug response
p.Leu858Arg
Osimertinib
Sensitivity
3 resistance signals
6 trials matched
CIViC therapy evidence
curated
ClinVar review status
expert panel
ClinicalTrials.gov matches
live search
Every claim resolves to a source
GRCh38
Illustrative case · synthetic data
47
candidate
matches
matches
01 / The workflow
The case is the unit of work.
Not a chat box. Not a variant list with a score glued on. A structured dossier that moves from raw call to board-ready report without losing the thread between a claim and its source.
-
01
Upload
Drop in a VCF or variant table. VEP 116 annotates against GRCh38, with ClinVar review status, gnomAD frequency, and normalized calls.
-
02
Review
Reportable variants are separated from incidental findings. Each call carries its provenance, so review starts with the right rows.
-
03
Weigh
Sensitivity, resistance, and off-label signals are matched with direction, level, and citation. Nothing is inferred without a source.
-
04
Ship
Generate a versioned PDF report and editable tumor-board deck. Finalize, share with an expiring link, and keep the audit trail.
02 / The evidence layer
Built for the way evidence review actually happens.
The hard part is not finding a drug name. It is knowing what the evidence says, how strong it is, and where it came from. Drugmarker keeps those questions attached to the case.
Resistance surfaced early
Resistance and acquired-resistance evidence sits beside the sensitivity signal, so the board sees the whole picture.
Citations on every claim
Source, evidence level, and direction travel with the match. Reviewers can open the trail instead of trusting a score.
Trial eligibility, not trial noise
ClinicalTrials.gov searches start from the case tumor type and reportable variants, then land in the same review surface.
De-identified by design
Cases are built around a patient code, not a chart identity. Keep research and clinical systems separate.
Reports with a memory
Versions, reviewers, share links, and finalization are recorded. The report is a document, not a screenshot.
03 / Governance
A research instrument, not a clinical decision.
Drugmarker is built for research and tumor-board preparation. It does not diagnose, recommend treatment, or replace a qualified clinician. The output is a dossier for expert review — not an answer key.
01
De-identified workflow
Patient codes, not names. Specimens and assays stay in the case record without becoming identity data.
02
Auditable review
Uploads, variant decisions, report versions, and share links leave a trail that can be reviewed later.
03
Single-tenant pilot
Early access runs in a bounded pilot. A clean path to multi-tenant isolation is part of the architecture.
RUO
Research use only. Not medical advice, not a diagnosis, not clinical
decision support. Verify all findings with a qualified clinician.
Pilot cohort
Bring one real workflow. We’ll bring the evidence layer.
Request pilot access and tell us what you are trying to review. We onboard research and clinical teams in small cohorts.