Geometric Protein Cavity Detection
Project information
- Start date: 25 May, 2020
- End date: Present
-
Geometric Protein Cavity Detection Cavities are often the locations of binding sites
that are paramount for protein interaction. This makes their automatic detection an
important step in drug discovery. Geometric Protein Cavity Detection (GPCD) is a
high-performance method for detecting protein cavities. The algorithm is designed to detect
most types of cavities, including clefts, invaginations, tunnels, channels, and voids. The
major advantage of GPCD over other cavity detection methods is that it can process large
proteins orders of magnitude faster. GPCD uses parallel image-to-mesh conversion to generate
an accurate protein surface and parallel ambient occlusion for detecting cavities on
heterogeneous nodes. This design can allow GPCD to quickly analyze protein systems
consisting of millions of atoms. GPCD's quality of results -- measured in terms of type and
number of cavities detected -- are comparable and in some cases better than existing
state-of-the-art algorithms for large proteins.