Calibrated Computational Triage of Approved Drugs against Mycobacterium tuberculosis InhA: Docking, Dynamics, and the Limits of Binding Inference

Direct inhibition of InhA, the enoyl-acyl carrier protein reductase targeted by isoniazid, could circumventKatG-dependent resistance, but computational repurposing must first show whether docking scores rankknown binders and whether post-docking physics adds information. We calibrated an AutoDock Vinaprotocol on 337 ChEMBL actives and 337 property-matched decoys, screened 2,147 approved smallmolecules against wild-type InhA (4DRE, NADH) and its S94A variant (4DTI, NADH), and tested theranking on the wild-type NAD+-bound structure 4U0J. The protocol enriched known actives (AUC 0.684and 0.713; EF5 1.59 and 1.65; permutation p = 0.001) and ranked prednisolone tebutate and entrectinibat the top of the approved library. In explicit-solvent MD, entrectinib retained the five key contacts overthe 5–20 ns window, as did the decoy, and gave an effective MM-GBSA estimate of −43.5 kcal mol−1,compared with −17.7 kcal mol−1 for the decoy; the 2 nM inhibitor GSK625 gave −33.1 kcal mol−1 as anorthogonal 5JFO/NAD+ control. A positive-control and scaffold-held-out audit showed that the top bandis a prioritization rather than a validated binder classifier, because strong inhibitors were not recoveredand held-out recovery was low. The resulting shortlist is, as a result, a calibrated hypothesis for enzymetesting

Authors

Publication Details

Journal
Figshare
Published
2026-11-18
DOI
https://doi.org/10.6084/m9.figshare.35048476
Primary Topic
Computational Drug Discovery Methods
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Calibrated Computational Triage of Approved Drugs against Mycobacterium tuberculosis InhA: Docking, Dynamics, and the Limits of Binding Inference

Zichun Qiu
1 links
Figshare
Computational Drug Discovery Methods
preprint

Calibrated Computational Triage of Approved Drugs against Mycobacterium tuberculosis InhA: Docking, Dynamics, and the Limits of Binding Inference

Zichun Qiu
preprint en

Abstract

Direct inhibition of InhA, the enoyl-acyl carrier protein reductase targeted by isoniazid, could circumventKatG-dependent resistance, but computational repurposing must first show whether docking scores rankknown binders and whether post-docking physics adds information. We calibrated an AutoDock Vinaprotocol on 337 ChEMBL actives and 337 property-matched decoys, screened 2,147 approved smallmolecules against wild-type InhA (4DRE, NADH) and its S94A variant (4DTI, NADH), and tested theranking on the wild-type NAD+-bound structure 4U0J. The protocol enriched known actives (AUC 0.684and 0.713; EF5 1.59 and 1.65; permutation p = 0.001) and ranked prednisolone tebutate and entrectinibat the top of the approved library. In explicit-solvent MD, entrectinib retained the five key contacts overthe 5–20 ns window, as did the decoy, and gave an effective MM-GBSA estimate of −43.5 kcal mol−1,compared with −17.7 kcal mol−1 for the decoy; the 2 nM inhibitor GSK625 gave −33.1 kcal mol−1 as anorthogonal 5JFO/NAD+ control. A positive-control and scaffold-held-out audit showed that the top bandis a prioritization rather than a validated binder classifier, because strong inhibitors were not recoveredand held-out recovery was low. The resulting shortlist is, as a result, a calibrated hypothesis for enzymetesting

Figshare
Good health and well-being
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