Genomarker AI drug-reversal benchmark: protocol v1

The frozen protocol of the Genomarker AI drug-reversal benchmark, deposited before any benchmark run. The benchmark asks how well a connectivity-style reversal ranking of LINCS L1000 compounds against each cancer's disease signature recovers the drugs approved for that cancer, compared with a leave-one-cancer-out frequent-hitter baseline and a proliferation-query baseline. protocol.yaml fixes the positives (approvals in the query cancer), the excluded trial stages, the minimum number of profiled positives, the recall fractions, the stratified bootstrap (2,000 draws, seed 20261111, 95% level), the validation rule and the normal-reference decision rule; its comment block states the tie and leave-out rules in plain words. sha256 of protocol.yaml: cad2b63a91784356df4f7272911a07be62cdfa5fe1830a610ff23e34a02953f2. Every benchmark run records this sha256 and this DOI. Nothing is tuned after a run; a method change is a new protocol version, published beside v1.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-15
DOI
https://doi.org/10.5281/zenodo.24264569
Primary Topic
Computational Drug Discovery Methods
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Genomarker AI drug-reversal benchmark: protocol v1

Genomarker AI
Zenodo (CERN European Organization for Nuclear Research)
Computational Drug Discovery Methods
article

Genomarker AI drug-reversal benchmark: protocol v1

Genomarker AI
article en

Abstract

The frozen protocol of the Genomarker AI drug-reversal benchmark, deposited before any benchmark run. The benchmark asks how well a connectivity-style reversal ranking of LINCS L1000 compounds against each cancer's disease signature recovers the drugs approved for that cancer, compared with a leave-one-cancer-out frequent-hitter baseline and a proliferation-query baseline. protocol.yaml fixes the positives (approvals in the query cancer), the excluded trial stages, the minimum number of profiled positives, the recall fractions, the stratified bootstrap (2,000 draws, seed 20261111, 95% level), the validation rule and the normal-reference decision rule; its comment block states the tie and leave-out rules in plain words. sha256 of protocol.yaml: cad2b63a91784356df4f7272911a07be62cdfa5fe1830a610ff23e34a02953f2. Every benchmark run records this sha256 and this DOI. Nothing is tuned after a run; a method change is a new protocol version, published beside v1.

Zenodo (CERN European Organization for Nuclear Research)
Computational Drug Discovery Methods
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon