Predicting the protein targets for athletic performance-enhancing substances

Abstract

BackgroundThe World Anti-Doping Agency (WADA) publishes the Prohibited List, a manually compiled international standard of substances and methods prohibited in-competition, out-of-competition and in particular sports. It would be ideal to be able to identify all substances that have one or more performance-enhancing pharmacological actions in an automated, fast and cost effective way. Here, we use experimental data derived from the ChEMBL database (~7,000,000 activity records for 1,300,000 compounds) to build a database model that takes into account both structure and experimental information, and use this database to predict both on-target and off-target interactions between these molecules and targets relevant to doping in sport.ResultsThe ChEMBL database was screened and eight well populated categories of activities (Ki, Kd, EC50, ED50, activity, potency, inhibition and IC50) were used for a rule-based filtering process to define the labels “active” or “inactive”. The “active” compounds for each of the ChEMBL families were thereby defined and these populated our bioactivity-based filtered families. A structure-based clustering step was subsequently performed in order to split families with more than one distinct chemical scaffold. This produced refined families, whose members share both a common chemical scaffold and bioactivity against a common target in ChEMBL.ConclusionsWe have used the Parzen-Rosenblatt machine learning approach to test whether compounds in ChEMBL can be correctly predicted to belong to their appropriate refined families. Validation tests using the refined families gave a significant increase in predictivity compared with the filtered or with the original families. Out of 61,660 queries in our Monte Carlo cross-validation, belonging to 19,639 refined families, 41,300 (66.98%) had the parent family as the top prediction and 53,797 (87.25%) had the parent family in the top four hits. Having thus validated our approach, we used it to identify the protein targets associated with the WADA prohibited classes. For compounds where we do not have experimental data, we use their computed patterns of interaction with protein targets to make predictions of bioactivity. We hope that other groups will test these predictions experimentally in the future.

Predicting targets of compounds against neurological diseases using cheminformatic methodology

“…In order to predict whether a given molecule will be active against a particular target, we first applied a number of rules on the ChEMBL dataset in order to generate sets of molecules that are experimentally determined to be bioactive [IC 50 (B50 lM), K i (\20 lM), K d (B10 lM), EC 50 (B40 lM), ED 50 (B40 lM), potency (B10 lM), activity (C40 %), inhibition (C45 %)]. These rules depend on the ranges of values against that target and the distribution of values of the relevant quantity within ChEMBL [56,58]. This process generates bioactivity based filtered families.…”

“…This leads to a set of refined families, each consisting of a group of molecules, which share similar chemical structure and bioactivity. The refined families of the ChEMBL dataset will allow us to identify the different sets of ligands [56,58,61].…”

“…The obtained Tanimoto similarity scores are then transformed into probabilities (pairwise p values) using an appropriate kernel function. The Gaussian distribution was proven to be the best suited kernel function for the refined ChEMBL dataset [58]. In order to predict molecule-target pairs, we had to calculate how similar a given molecule x i is to the members of family x = {x 1 , x 2 , …, x n } using the refined ChEMBL dataset.…”

“…This process generates bioactivity based filtered families. Our recently developed PFClust clustering [61] was applied to all the filtered ChEMBL families, which subdivided each family into smaller groups based both on ligand structure and their proven activity on a given protein target [56,58]. The compounds were clustered on the basis of their chemical structures, described by circular fingerprints (CFPs) [57].…”

“…Our interest is in predicting ligand-target associations that will allow us to define binding profile of ligand and in suggesting theoretical and experimental approaches directed towards gaining a deeper understanding of possible pharmacological effects. Our novel methodology, based on circular fingerprint (CFP) descriptors of compounds [57] and information data mined in the ChEMBL database, was very successfully applied in prediction of unexplored compound-to-target associations using a set of the WADA prohibited compounds [58]. A similar approach [58] was now applied to determine primary pharmaceutical targets and off-targets for our novel multi-target ligands (1-134) [23][24][25][26][27][28][29][30][31] as a crucial step in understanding the pharmacological and toxicological profiles of these novel compounds.…”