A Comparison between Enrichment Optimization Algorithm (EOA)-Based and Docking-Based Virtual Screening

Abstract

Virtual screening (VS) is a well-established method in the initial stages of many drug and material design projects. VS is typically performed using structure-based approaches such as molecular docking, or various ligand-based approaches. Most docking tools were designed to be as global as possible, and consequently only require knowledge on the 3D structure of the biotarget. In contrast, many ligand-based approaches (e.g., 3D-QSAR and pharmacophore) require prior development of project-specific predictive models. Depending on the type of model (e.g., classification or regression), predictive ability is typically evaluated using metrics of performance on either the training set (e.g.,QCV2) or the test set (e.g., specificity, selectivity or QF1/F2/F32). However, none of these metrics were developed with VS in mind, and consequently, their ability to reliably assess the performances of a model in the context of VS is at best limited. With this in mind we have recently reported the development of the enrichment optimization algorithm (EOA). EOA derives QSAR models in the form of multiple linear regression (MLR) equations for VS by optimizing an enrichment-based metric in the space of the descriptors. Here we present an improved version of the algorithm which better handles active compounds and which also takes into account information on inactive (either known inactive or decoy) compounds. We compared the improved EOA in small-scale VS experiments with three common docking tools, namely, Glide-SP, GOLD and AutoDock Vina, employing five molecular targets (acetylcholinesterase, human immunodeficiency virus type 1 protease, MAP kinase p38 alpha, urokinase-type plasminogen activator, and trypsin I). We found that EOA consistently outperformed all docking tools in terms of the area under the ROC curve (AUC) and EF1% metrics that measured the overall and initial success of the VS process, respectively. This was the case when the docking metrics were calculated based on a consensus approach and when they were calculated based on two different sets of single crystal structures. Finally, we propose that EOA could be combined with molecular docking to derive target-specific scoring functions.

Towards an Enrichment Optimization Algorithm (EOA)‐based Target Specific Docking Functions for Virtual Screening

“…We attribute this decrease to the much smaller percentage of active compounds in these sets (see Method section) making them more difficult to identify within a large pool of decoys. We have previously observed a similar decrease in EOA performances for five other protein targets [43] . Of note the performances of these EOA models are poorer than the one previously reported by us [28,43] .…”

“…We have previously observed a similar decrease in EOA performances for five other protein targets [43] . Of note the performances of these EOA models are poorer than the one previously reported by us [28,43] . We attribute this to the much smaller number of descriptors used for the derivation of the EOA models.…”

“…The performances of the EOA models on training, validation and test sets in terms of the number of retrieved active compounds are reported in Table 2. These are lower than those previously reported by us on other datasets [28,43] . We attribute this to the much smaller number of descriptors used in the derivation of the EOA equations (4–5 descriptors) in comparison with our previous works, as well as to our focusing on “difficult” targets (see above).…”

“…In our previous works,[ 28 , 43 ] we presented a novel algorithm for the derivation of multiple linear regression (MLR) equations suitable for usage in virtual screening, based on the optimization of an enrichment‐aware objective function. We termed the new algorithm, enrichment optimizer algorithm (EOA).…”

“…As noted above, all the components that make up GOLD's ChemPLP scoring function were used as descriptors except for the Smetal descriptor for all datasets and the Scho descriptor for the PGH1 and CYP2C9 datasets. In contrast to our previous works, [28,43] in the present study EOA models were derived using all descriptors and only the descriptors weights were optimized. Furthermore, weights were allowed to vary only within a limited range around their original values.…”