How does consensus scoring work for virtual library screening? An idealized computer experiment

It has been reported recently that consensus scoring, which combines multip le scoring functions in binding affinity estimation, leads to. higher hit-r ates in virtual library screening studies. This method seems quite independ ent to the target receptor, the docking program, or even the scoring functi ons under investigation. Here we present an idealized computer experiment t o explore how consensus scoring works. A hypothetical set of 5000 compounds is used to represent a chemical library under screening. The binding affin ities of all its member compounds are assigned by mimicking a real situatio n. Based on the assumption that the error of a scoring function is a random number in a normal distribution, the predicted binding affinities were gen erated by adding such a random number to the "observed" binding affinities. The relationship between the hit-rates and the number of scoring functions employed in scoring was then investigated. The performance of several typi cal ranking strategies for a consensus scoring procedure was also explored. Our results demonstrate that consensus scoring outperforms any single scor ing for a simple statistical reason: the mean value of repeated samplings t ends to be closer to the true value. Our results also suggest that a modera te number of scoring functions, three or four, are sufficient for the purpo se of consensus scoring. As for the ranking strategy, both the rank-by-numb er and the rank-by-rank strategy work more effectively than the rank-by-vot e strategy.