Prioritization of high throughput screening data of compound mixtures using molecular similarity

Abstract

This work addresses the problem of enriching high throughput screening (HTS) data of mixtures of five compounds/well without any redundancies and where no structural information of the drug target is available. Devising a mathematical model for mixtures of compounds is extremely difficult. Instead, we employed extended-connectivity fingerprints (ECFPs), which are a new class of fingerprints for molecular characterization. We calculated the Tanimoto similarity between all the compounds which appeared in the active mixtures and ranked them in order of decreasing similarity. This methodology enriched the data in three recent in-house HTS campaigns that were conducted in five compounds/well mixtures. The top 10% most similar compounds captured between 29% and 41% of the active compounds. Although this methodology is not particularly sensitive, considering the quality of the data (ca. 80% noise presumed to be inactive) and the simplicity of the method, we find it useful. It offers a true opportunity to quickly prioritize and enrich primary screening data from mixtures of compounds and therefore reduce the time and the high costs of the secondary screening.