AutoDock-SS: AutoDock for Multiconformational Ligand-Based Virtual Screening

Ligand-based virtual screening (LBVS) can be pivotal for identifying potential drug leads, especially when the target protein’s structure is unknown. However, current LBVS methods are limited in their ability to consider the ligand conformational flexibility. This study presents AutoDock-SS (Similarity Searching), which adapts protein–ligand docking for use in LBVS. AutoDock-SS integrates novel ligand-based grid maps and AutoDock-GPU into a novel three-dimensional LBVS workflow. Unlike other approaches based on pregenerated conformer libraries, AutoDock-SS’s built-in conformational search optimizes conformations dynamically based on the reference ligand, thus providing a more accurate representation of relevant ligand conformations. AutoDock-SS supports two modes: single and multiple ligand queries, allowing for the seamless consideration of multiple reference ligands. When tested on the Directory of Useful Decoys—Enhanced (DUD-E) data set, AutoDock-SS surpassed alternative 3D LBVS methods, achieving a mean AUROC of 0.775 and an EF1% of 25.72 in single-reference mode. The multireference mode, evaluated on the augmented DUD-E+ data set, demonstrated superior accuracy with a mean AUROC of 0.843 and an EF1% of 34.59. This enhanced performance underscores AutoDock-SS’s ability to treat compounds as conformationally flexible while considering the ligand’s shape, pharmacophore, and electrostatic potential, expanding the potential of LBVS methods.


Relationship Between Degrees of Freedom and AutoDock-SS Performance: An Analytical Investigation
In our study, we further investigated the impact of the Degree of Freedom (DOF) of library compounds on the performance of AutoDock-SS, utilizing the number of rotatable bonds (NRotB) as a proxy for DOF.Initially, we examined the relationship between the mean and median values of NRotB within the actives and decoys libraries of 102 DUD-E targets and their corresponding AUROC scores.The Pearson correlation coefficients obtained were 0.26 and 0.31 respectively, with p-values equal to 0.02 and 0.00, respectively, indicating a weak positive correlation between both mean and median values of NRotB and AUROC.
Subsequently, a more detailed experiment was conducted.For each target, library compounds were categorized into groups based on the number of NRotB: 0-4, 5-8, 9-12, and over 12.The ratio of actives and decoys in each group was adjusted while maintaining the original ratio of the full actives to decoys libraries using a random selection process.New AUROC values were then recalculated for each group.The compiled results across all DUD-E targets were as follows:  NRotB 0-4: Mean AUROC of 0.

Figure S1 .
Figure S1.Comparative Analysis of Performance Metrics for AutoDock-SS Single-reference and Multi-reference Modes.A) Illustration of the AUROC values for both modes, with the x-axis representing the AUROC for the Single-reference mode and the y-axis for the Multi-reference mode.Each point represents an individual measurement, and the dashed line indicates the line of equality where the performance of both modes would be identical.B) Illustration of the EF 1% value for the two modes, with the Single-reference mode on the x-axis and the Multi-reference mode on the y-axis.Similar to Panel A, each point signifies an individual data point, and the dashed line represents identical performance across both modes.Data points above the dashed line in both panels suggest a performance advantage for the Multi-reference mode over the Single-reference mode.

Figure S2 .
Figure S2.The cluster of 92 ROCs from AutoDock-SS multi-reference mode screening of all DUD-E + targets.The orange boxplot on the right represents the distribution of AUC values, and the blue one shows the distribution of EF1% values.The red line in the boxplot indicates the median AUROC value, and the green triangle represents the mean AUROC value.

Table S1
Atom types required by affinity maps for AutoDock-SS If the AUROC value difference between the two modes is greater than 0.100, the higher AUROC value is bold.
NRotB 9-12: Mean AUROC of 0.48, Std of 0.15  NRotB Over 12: Mean AUROC of 0.41, Std of 0.23 An Analysis of Variance (ANOVA) test conducted on these results yielded an F-value of 7.82 with a p-value equaled to 0.00.This F-value indicates that there is significant variation among different torsion ranges.It was observed that AutoDock-SS achieved better performance in groups with a lower number of NRotB, and a slight decrease in performance was noted with increasing NRotB.S3.Distribution of AUROC values across different ranges of rotatable bonds.This boxplot displays the variability and central tendency of AUROC values for compounds categorized by the number of rotatable bonds.Outliers indicate compoundswith AUROC scores that are notably distinct from the typical score range in each category.