Skip to Main Content
We developed a new method to improve the accuracy of molecular interaction data using a protein-compound affinity matrix calculated by protein-compound docking software. We approximated the protein-compound binding free energy as a linear combination of the raw docking scores of the compound with many different proteins. The coefficients of the linear combination were estimated based on the amino-acid sequence similarities among proteins. This method was applied to in silico screening of the active compounds of five target proteins using multiple target screening, and it increased the hit ratio by several times compared to that given by the raw docking scores. The hit ratio also becomes robust against the difference of target proteins. In addition, we have developed some methods based on a protein-compound affinity matrix. When some active compounds were known, a consensus score, which combines the structure-based and ligand-based screening results, was applied to a target. Finally, we could achieve a high hit ratio for some targets by using a combination of screening methods.