ProTailor: a Parallel Operator for Extremely Fast Shape Analysis in Bioinformatics Applications

The geometric shape of molecular surfaces strongly influences the docking processes where, although electrostatic, hydrophobic and van der Waals interactions affect greatly the binding affinity of the molecules, shape complementarity is a necessary condition. The vast majority of molecular docking algorithms uses a brute force enumeration of the transformation space, which requires extremely long running times. Few other methods use local shape feature matching to reduce the search to those relative positions which satisfy geometric constraints. Based on a shape analysis tool developed in Computer Graphics, in this paper we introduce ProTailor, a parallel algorithm for efficient multi-scale detection of morphological features on molecular surfaces. Thanks to an almost linear speed-up, we show how ProTailor is well suited to efficiently identify salient features like cavities and depressions, saddle areas or bridges. Feature identification may serve as a powerful tool to automatically locate potential binding sites or as a pre-processing step for efficient shape complementarity assessment in docking prediction.