Ring-Conformer Ensembles Improve AutoDock Vina Docking; an Ant Colony Makes Them Affordable
Abstract
In standard preparation, AutoDock Vina holds five-and six-membered non-aromatic rings at their input pucker, and our failure analysis of 4 509 runs against searchreachable native poses shows what that costs: with the accurate preset, the search itself fails in only 6.5% of runs, while ring conformations, ranking and the scoring function account for most of the rest. Docking up to four ring conformers per ligand and keeping the best-scoring pose repairs part of that gap for every search we tested, raising top-1 success by 5.5 to 7.4 percentage points for Vina’s Monte Carlo at exhaustiveness 1 and 8 alike and for the two presets we tested it with in the continuous ant colony optimization (ACOR) search we built into Vina; a cost-matched control that docks the input conformer K times shows that most of this gain is ring geometry rather than repeated sampling (+8.9 against +2.9 points with the accurate preset). Because the ensemble multiplies the cost of docking, its price is what limits its use, and this is where the ant colony earns its place: it reaches the pose accuracy of Vina at exhaustiveness 1 with about nine times fewer scoring evaluations, so an ensemble of up to four conformers with the accurate preset costs 8.5 s per ligand on one core and reaches 52.7% top-1 success, against 50.4% for Vina at exhaustiveness 8 (37.9 s) and 46.1% for the same ensemble driven by exhaustiveness 1. Ring ensembles with Vina’s own Monte Carlo at exhaustiveness 8 are the most accurate arm we measured (56.2%), at thirteen times that cost, so the choice between them is one of budget rather than of search.
Related
- A Congener-Substitution Strategy for Docking Heavy Metal Protein-Ligand Complexes with AutoDock Vina
- Parity-equivariant prediction of conformer-resolved electronic circular dichroism spectra accelerates absolute configuration assignment
- Conformer diversity over quantity in COSMO-RS-based LLE predictions
- CRAFT-ε: A Conformer-Resolved Atom- and Fragment-based Theory for the Static Relative Permittivity of Pure Organic Liquids
- Site-preference engineering of cobalt active-site ensembles in spinel oxides for oxygen evolution
BibTeX
@misc{cinaroglu2026ringconformer,
title = {{Ring-Conformer Ensembles Improve AutoDock Vina Docking; an Ant Colony Makes Them Affordable}},
author = {Suleyman Selim Cinaroglu},
howpublished = {ChemRxiv},
year = {2026},
month = sep,
doi = {10.26434/chemrxiv.15009672/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15009672/v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Ring-Conformer Ensembles Improve AutoDock Vina Docking; an Ant Colony Makes Them Affordable
AU - Suleyman Selim Cinaroglu
PY - 2026
DA - 2026/09/29
PB - ChemRxiv
DO - 10.26434/chemrxiv.15009672/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15009672/v1
AB - In standard preparation, AutoDock Vina holds five-and six-membered non-aromatic rings at their input pucker, and our failure analysis of 4 509 runs against searchreachable native poses shows what that costs: with the accurate preset, the search itself fails in only 6.5% of runs, while ring conformations, ranking and the scoring function account for most of the rest. Docking up to four ring conformers per ligand and keeping the best-scoring pose repairs part of that gap for every search we tested, raising top-1 success by 5.5 to 7.4 percentage points for Vina’s Monte Carlo at exhaustiveness 1 and 8 alike and for the two presets we tested it with in the continuous ant colony optimization (ACOR) search we built into Vina; a cost-matched control that docks the input conformer K times shows that most of this gain is ring geometry rather than repeated sampling (+8.9 against +2.9 points with the accurate preset). Because the ensemble multiplies the cost of docking, its price is what limits its use, and this is where the ant colony earns its place: it reaches the pose accuracy of Vina at exhaustiveness 1 with about nine times fewer scoring evaluations, so an ensemble of up to four conformers with the accurate preset costs 8.5 s per ligand on one core and reaches 52.7% top-1 success, against 50.4% for Vina at exhaustiveness 8 (37.9 s) and 46.1% for the same ensemble driven by exhaustiveness 1. Ring ensembles with Vina’s own Monte Carlo at exhaustiveness 8 are the most accurate arm we measured (56.2%), at thirteen times that cost, so the choice between them is one of budget rather than of search.
ER -