Improving the Accuracy of an Affinity Prediction Method by Using Statistics on Shape Complementarity between Proteins

Abstract

To elucidate the partners in protein-protein interactions (PPIs), we previously proposed an affinity prediction method called affinity evaluation and prediction (AEP), which is based on the shape complementarity characteristics between proteins. The structures of the protein complexes obtained in our shape complementarity evaluation were selected by a newly developed clustering method called grouping. Our previous experiments showed that AEP gave accuracies that differed with the data composition and scale. In this study, we set a data scale (84 x 84 = 7056 protein pairs) including 84 biologically relevant complexes and then designed 225 parameter sets based on four key parameters related to the grouping and the calculation of affinity scores. As a result of receiver operating characteristic analysis, we obtained 27.4% sensitivity (= recall), 91.0% specificity, 3.5% precision, 90.2% accuracy, 6.3% F-measure(max), and an area under the curve of 0.585. Chiefly by optimization of the grouping, AEP was able to provide prediction accuracy for a maximum F-measure that statistically distinguished 23 target complexes among 84 protein pairs. Moreover, the active sites of these complexes were successfully predicted with high accuracy (i.e., 2.37 angstroms in 1CGI and 2.38 angstroms in 1PPE) of interface RMSD. To assess the improvement in accuracy we compared the results of AEP of different data sets and of tentative methods using ZDOCK 3.0.1 or ZRANK scores.

IN SILICO SCREENING OF PROTEIN–PROTEIN INTERACTIONS WITH ALL-TO-ALL RIGID DOCKING AND CLUSTERING: AN APPLICATION TO PATHWAY ANALYSIS

“…28 For comparison we applied our system to the entire dataset of the proteinprotein docking benchmark 2.0 (84 × 84 combinations) as in the previous study. 18 We obtained a best F-measure value of 0.33 with m * = 2.0, E * = 8.6, which outperformed the referenced work (F-measure value of 0.063). Other assessment values of sensitivity ( T P T P +F N ), specificity ( T N F P +T N ) , precision ( T P T P +F P ) and accuracy ( There has been a lot of effort expended to identify the PPI network and protein interacting sites by integrating genomic, transcriptomic and proteomic data.…”

“…Our method is similar to those in previous studies of Tsukamoto et al 17 and Yoshikawa et al; 18 however, in this study we used ZDOCK 3.0.1 for docking and different criteria for the clustering analysis and PPI detection. In the previous studies, the optimized parameters were different from those obtained here.…”

“…It is also notable that the range of the parameter set that resulted in good F-measure values was consistent between the two datasets. The method was also applied to the same dataset that was used in a previous related study 18 and outperformed it with a higher F-measure value of 0.33, compared to 0.063 in the earlier study.…”

“…Their method (Affinity Evaluation and Prediction, AEP) utilizes protein docking to find potentially interacting protein pairs as well as to indicate correct binding modes. 17,18 AEP consists of a two-stage analysis. Firstly, all combinations of protein structures are considered for all-to-all docking based on shape complementarity.…”

Challenges and advances in computational docking: 2009 in review

“…This review is not limited to, but is strongly focused on docking advances in the context of drug design, specifically in the areas of virtual screening (VS) (1) and fragment-based drug design (FBDD). The area of protein-protein docking will not be covered in this review (but references are provided here for readers' convenience) (2)(3)(4)(5)(6)(7)(8)(9)(10)(11). Miscellaneous case studies, except where important methodological developments are described, are also beyond the scope of this review.…”