Large scale meta-analysis of fragment-based screening campaigns: privileged fragments and complementary technologies.

A first step in fragment-based drug discovery (FBDD) often entails a fragment-based screen (FBS) to identify fragment"hits." However, the integration of conflicting results from orthogonal screens remains a challenge. Here we present a meta-analysis of 35 fragment-based campaigns at Novartis, which employed a generic 1400- fragmentlibrary against diverse target families using various biophysicaland biochemical techniques. By statistically interrogating the multidimensional FBS data, we sought to investigate three questions: (1) What makes a fragmentamenable for FBS? (2) How do hits from different fragmentscreening technologies and target classes compare with each other? (3) What is the best way to pair FBS assay technologies? In doing so, we identified substructures that were privileged for specific target classes, as well as fragmentsthat were privileged for authentic activity against many targets. We also revealed some of the discrepancies between technologies. Finally, we uncovered a simple ruleof thumbin screening strategy: when choosing two technologies for a campaign, pairing a biochemical and biophysicalscreen tends to yield the greatest coverage of authentic hits.