A Heuristic Algorithm for Plate SelectionThat Maximizes Compound Diversity

Abstract

Abstract. A heuristic algorithm was developed to maximize compound diversity within a subset of screening plates used in high throughput screening (HTS) to initiate the drug discovery process. The approach overcame the challenge of combinatorial explosion for selecting plate subsets with maximum compound diversity. The method yielded novel forms of plate-based diversity subsets (PBDS) for HTS screening in lead discovery. The algorithm, its validation and the application to our screening collection are outlined.

Plate-based diversity subset screening generation 2: an improved paradigm for high-throughput screening of large compound files

“…Each screening plate contains up to 360 test compounds and has the remaining 24 positions reserved for control compounds. The Ro40 [ 35 ] specifies that in order to be compliant, a 384-well plate must have more than 200 unique test compounds, more than 160 test compounds with zero Ro5 [ 39 ] violations and more than 120 test compounds free of any undesirable structural alerts [ 40 ] These rules were modified and made stricter for PBDS2 plate selection as follows (see Experimental section for details): remove plates with very low numbers (<200) and penalize plates with low numbers (<300) of unique compounds passing the GDRS filters; remove plates with very high numbers (>300) and penalize plates with high numbers (>160) of total Ro5 violations, summed up across the plate; remove plates with very high numbers (>100) and penalize plates with high numbers (>30) of compounds failing the stricter structural filters [ 35 , 40 ]; give combinatorial plates an initial advantage in the selection process but remove this advantage in the final optimization iterations to achieve convergence (see below); remove all plates already in PBDS in order to avoid plate depletion issues [ 41 ]. Experience from constructing the PBDS subset [ 35 ], which was very successful in practice, had shown that plates in violation of the Ro40 which were severely penalized (their assigned random number received a penalty of ) in the seeding process never got selected.…”

“…The performance of the PBDS2 was excellent, generating high-quality hit series that translated to physiologically relevant systems and formed the basis for medicinal chemistry design and synthesis [ 51 , 53 ]. Demand for access to this newly prioritized plate-based diverse subset was so great that eventually the material management stocks, which had been designed to last for ten years, became depleted and a complementary subset approach PBDSx was subsequently introduced to mitigate this [ 41 ].…”

“…Indeed, a new approach to the plate selection algorithm has recently emerged from Novartis, where plates are selected principally according to the biological target diversity profile of the compounds on the plates [ 43 ]. A new variant has also emerged from Pfizer, where, due to increased use of PBDS2 and differential plate depletion, a novel methodology for the construction of several equivalent PBDSs was developed [ 41 ]. We look forward to further developments to improve drug discovery efficiency in the future.…”

Performance of Dark Chemical Matter in High Throughput Screening

“…Screening pools at pharmaceutical companies are being revisited improving the quality of compounds using analytical methods and also eliminating redundancies in chemical space . Other strategies of reducing the number of compounds to be screened include the stratification of screening compounds allowing for diversity-based or druglikeness based selections of screening plates for subscreens. − …”

The potential use of single-particle electron microscopy as a tool for structure-based inhibitor design

“…These large libraries cover only a small fraction of chemical space, which may hinder the discovery of inhibitors for targets with unusual binding sites such as allosteric sites or protein-protein interactions (PPIs). To reduce the cost of HTS, new strategies involving smaller numbers of diverse compounds are being sought (Crisman et al, 2007;Zhu et al, 2013;Nissink et al, 2014).…”