Scoring Ligand Efficiency: Potency, Ligand Efficiency and Product Ligand Efficiency within Big Data Landscape

Abstract

Background: Potency is the broadest available biological activity data type. In turn, Ligand Efficiency (LE) is a molecular descriptor that probes the ratio of potency vs Heavy Atom Count (HAC), which emphasizes low HAC more than potency and thus has drawbacks as an estimator of drug candidates. The objective was to design a novel transform to probe potency and HAC interaction in which potency and HAC would be balanced more evenly. Methods: In this study, potency data of ChEMBL, PubChem, FDA approvals and drug (fragments) were analysed. A novel descriptor, a product of the pAC50 value with HAC, multiplicative or Product Ligand Efficiency (PLE) was designed and tested. Results: In particular PLE was compared with pAC50 and LE vs the HAC statistics for different series of ligands. This indicated that PLE is an informative estimator that can be used to recognize the potential of drugs. PLE has a maximum value in the range around 30-50 HAC. Conclusion: Drug design is a complex problem. Similarly, to drug-likeness, LE prefers small molecules. This makes LE a tool serendipitously improving drug likeness. In this context, LE performs unexpectedly well even despite the uncertainty of its physical meaning. PLE is a more evenly balanced estimator whose physical meaning is the Minimum Inhibitory Concentration (MIC).

A Brief Guide to Big Data in Molecular Design: From Concepts and Definitions to Models

“…In other words, notso-big data is a pragmatic approach replacing big data due to better interpretability or simply when we cannot expand the data space. Chemo-, pharmaco-, pharmaco-economic substance behavior analyses were published using large data available [22][23][24][25]. In molecular design, we can suggest the term all-data available.…”