Substituting random forest for multiple linear regression improves binding affinity prediction of scoring functions: cyscore as a case study (BMC Bioinformatics)

Background: State-of-the-art protein-ligand docking methods are generally limited by the traditionally lowaccuracy of their scoring functions, which are used to predict binding affinity and thus vital fordiscriminating between active and inactive compounds. Despite intensive research over the years,classical scoring functions have reached a plateau in their predictive performance. These assume apredetermined additive functional form for some sophisticated numerical features, and use standardmultivariate linear regression (MLR) on experimental data to derive the coefficients. Results: In this study we show that such a simple functional form is detrimental for the predictionperformance of a scoring function, and replacing linear regression by machine learning techniqueslike random forest (RF) can improve prediction performance. We investigate the conditions ofapplying RF under various contexts and find that given sufficient training samples RF manages tocomprehensively capture the non-linearity between structural features and measured bindingaffinities. Incorporating more structural features and training with more samples can both boost RFperformance. In addition, we analyze the importance of structural features to binding affinityprediction using the RF variable importance tool. Lastly, we use Cyscore, a top performing empiricalscoring function, as a baseline for comparison study. Conclusions: Machine-learning scoring functions are fundamentally different from classical scoring functionsbecause the former circumvents the fixed functional form relating structural features with bindingaffinities. RF, but not MLR, can effectively exploit more structural features and more trainingsamples, leading to higher prediction performance. The future availability of more X-ray crystalstructures will further widen the performance gap between RF-based and MLR-based scoringfunctions. This further stresses the importance of substituting RF for MLR in scoring functiondevelopment.