Pharmaceutical Machine Learning: Virtual High-Throughput Screens Identifying Promising and Economical Small Molecule Inhibitors of Complement Factor C1s

Abstract

When excessively activated, C1 is insufficiently regulated, which results in tissue damage. Such tissue damage causes the complement system to become further activated to remove the resulting tissue damage, and a vicious cycle of activation/tissue damage occurs. Current Food and Drug Administration approved treatments include supplemental recombinant C1 inhibitor, but these are extremely costly and a more economical solution is desired. In our work, we have utilized an existing data set of 136 compounds that have been previously tested for activity against C1. Using these compounds and the activity data, we have created models using principal component analysis, genetic algorithm, and support vector machine approaches to characterize activity. The models were then utilized to virtually screen the 72 million compound PubChem repository. This first round of virtual high-throughput screening identified many economical and promising inhibitor candidates, a subset of which was tested to validate their biological activity. These results were used to retrain the models and rescreen PubChem in a second round vHTS. Hit rates for the first round vHTS were 57%, while hit rates for the second round vHTS were 50%. Additional structure–property analysis was performed on the active and inactive compounds to identify interesting scaffolds for further investigation.

Identifying de‐NEDDylation inhibitors: Virtual high‐throughput screens targeting SENP8

“…Based on experimental validation results and goals, models are retrained using a training set comprised of both the original data and the newly obtained validation data. Another round of vHTS and experimental validation is conducted to determine if the retrained models made more accurate predictions, a trend observed in previous work (Chen & Visco, , ; Chen et al., ).…”

“…The work presented here is one part in a series (Chen, Schmucker, & Visco, ; Chen & Visco, , ) aiming to systematically determine the effectiveness of the approach in different systems and datasets of varying size and active/inactive class proportions. Previous work includes the identification of inhibitors for clotting factor XIIa (Chen & Visco, ), cathepsin L (Chen & Visco, ), and complement factor C1s (Chen et al., ). SENP8 was chosen because of the implicated systems it has control over.…”

Design and Selection of Novel C1s Inhibitors by In Silico and In Vitro Approaches

“…Due to poor pharmacokinetics, introduction of pegylated linkers led to related compounds with good potency and in vivo pharmacokinetic properties [8]. Recently, various inhibitors lacking the amidine (guanidine) “warheads” were identified by HTS technologies (Molecular Library Screening Center Network (MLSCN); Penn Center for Molecular Discovery (PCMD); listed in PubChem) [9,10] and reported by Chen et al [11]. Due to the limited availability of the amidine and guanidine derivatives from commercial repositories we have chosen the reported O, N-heterocycles (lacking the “warheads”) and the potential bioisosteric replacements of amidine (guanidine) motifs as starting points for selecting C1s focused compound libraries.…”

Machine Learning for Molecular Modelling in Drug Design

“…Chen et al [9] aimed at identifying new inhibitors of the C1 target that could be used to advance towards new treatments for hereditary angioedema. The QSAR models were built integrating SVM with PCA and GA-based FS.…”

“…Machine learning has been used to generate diverse ligand-based predictive models in these four contributions so far [8,9,10,11] by exploiting chemical structure and bioactivity data. However, by also exploiting X-ray crystal structure data, ML can also be used to build protein-ligand predictive models.…”