Deriving Knowledge through Data Mining High-Throughput Screening Data

Abstract

Deriving general knowledge from high-throughput screening data is made difficult by the significant amount of noise, arising primarily from false positives, in the data. The paradigm established for screening an encoded combinatorial library on polymeric support, an ECLiPS library, has a significant amount of built-in redundancy. Because of this redundancy, the resulting data can be interpreted through a rigorous statistical analysis procedure, thereby significantly reducing the number of false positives. Here, we develop the statistical models used to analyze data from high-throughput screens of ECLiPS libraries to derive unbiased true hit rates. These hit rates can also be calculated on subsets of the collection such as those compounds containing a carboxylic acid or those with molecular weight below 350 Da. The relative value of the hit rate on the subset of the collection can then be compared to the overall hit rate to determine the effect of the substructure or physical property on the likelihood of a molecule having biological activity. Here, we show the effects that various functional groups and the standard physical properties, molecular weight, hydrogen bond donors, hydrogen bond acceptors, log P, and rotatable bonds, have on the likelihood of a compound being biologically active. To our knowledge this is the first published account of the use of high-throughput screening data to elucidate the effects of physical properties and substructures on the likelihood of compounds showing biological activity over a broad range of pharmaceutically relevant targets.

Data mining PubChem using a support vector machine with the Signature molecular descriptor: Classification of factor XIa inhibitors

“…For example, if only 2% of the compounds were active, classifying all compounds as inactive would yield an accuracy of 98%, but not be a useful model. The second issue has to do with the quality of data obtained from HTS being noisy primarily from false positives [51]. For these reasons we decided to train on the confirmatory assay (AID 846) because it is relatively balanced and false positives are not as likely because the IC 50 determination was performed in triplicate.…”

Role of Chemistry in Lead Discovery

“…In a less cited article, Diller and Hobbs provide further support for the greater frequency of compounds with lead-like properties turning up as actives in HT screens [16]. The data was mined from the screening results obtained from 2 Â 10 8 compounds assayed across 100 molecular targets.…”

“…Pardridge: blood brain barrier permeability (CNS penetration) [9] HBD + HBA o 8; MW o 400; no acids Lipinski: Ro5 (oral drug-likeness) [10] cLogP W À2 to o5; MW o 500; HBD r 5; HDA r 10; rotatable bonds r10 Kelder: polar surface area (PSA) oral bioavailability [11] o120 Å (all drug types); o60-70 Å (CNS active) Clark: blood brain barrier permeability (CNS penetration) [12,13] [14,15] MW 100-350; cLogP 1-3 Veber: oral bioavailability (rat) [17] r10 rotatable bonds, r140 Å PSA, or o12 HBD + HBA independent of MW Diller: parameters for highest hit rates in large lead discovery libraries [16] cLogP 2-6; HBD 2; rotatable bonds 6; polar substituents able to make strong intermolecular contacts; r1 amide bond 9.2. DESIGN CONSIDERATIONS generally show larger relative hit rates ( Fig.…”

On the Relationship between Molecular Hit Rates in High-Throughput Screening and Molecular Descriptors

“…[6][7][8][9][10] Hopkins et al 6 reported that compound promiscuity decreases with higher molecular weight (MW) based on HTS data. Diller and Hobbs 11 published a statistical analysis of a company's historical HTS data, in which they examined the relationship between physicochemical properties, substructures, and likelihood of displaying biological activity, irrespective of target class.…”