Benchmark of four popular virtual screening programs: construction of the active/decoy dataset remains a major determinant of measured performance

Benchmark of four popular virtual screening programs: construction of the active/decoy dataset remains a major determinant of measured performance