Model for high-throughput screening of drug immunotoxicity – Study of the anti-microbial G1 over peritoneal macrophages using flow cytometry

Abstract

Quantitative Structure-Activity (mt-QSAR) techniques may become an important tool for prediction of cytotoxicity and High-throughput Screening (HTS) of drugs to rationalize drug discovery process. In this work, we train and validate by the first time mt-QSAR model using TOPS-MODE approach to calculate drug molecular descriptors and Linear Discriminant Analysis (LDA) function. This model correctly classifies 8258 out of 9000 (Accuracy = 91.76%) multiplexing assay endpoints of 7903 drugs (including both train and validation series). Each endpoint correspond to one out of 1418 assays, 36 molecular and cellular targets, 46 standard type measures, in two possible organisms (human and mouse). After that, we determined experimentally, by the first time, the values of EC50 = 21.58 μg/mL and Cytotoxicity = 23.6% for the anti-microbial/anti-parasite drug G1 over Balb/C mouse peritoneal macrophages using flow cytometry. In addition, the model predicts for G1 only 7 positive endpoints out 1251 cytotoxicity assays (0.56% of probability of cytotoxicity in multiple assays). The results obtained complement the toxicological studies of this important drug. This work adds a new tool to the existing pool of few methods useful for multi-target HTS of ChEMBL and other libraries of compounds towards drug discovery.

Exploring the role of quantum chemical descriptors in modeling acute toxicity of diverse chemicals to Daphnia magna

“…However, there have been equally important advanced and sophisticated QSAR approaches based on the classification method [13] which can effectively predict whether a chemical is biologically active or inactive. Recently [14][15][16][17][18][19][20][21], such classification based QSAR models have been reported for predicting the toxicity, of large and heterogeneous datasets of compounds, against many organisms, besides assessing multiple toxicological profiles under diverse experimental conditions. For example, Tenorio-Borroto et al [14][15][16], had proposed multi-target quantitative structure-activity/property relationships (mt-QSAR/QSPR) models along with the flow cytometry analysis for the prediction of cytotoxicity and immunotoxicity, which can effectively models the drug-target interactions and effects of organic compounds over the cellular and molecular targets of immune system.…”

Enabling the Discovery and Virtual Screening of Potent and Safe Antimicrobial Peptides. Simultaneous Prediction of Antibacterial Activity and Cytotoxicity

“…47 Thus, in numerous fields of research in drug discovery, various works have reported the development of advanced chemoinformatic models inspired by the idea regarding the calculation of Box-Jenkins moving averages. [33][34][35][36][48][49][50][51][52][53][54][55][56][57][58] In the first step, the following equation is employed: In Eq. 3, n(c r ) represents the number of peptides assayed by considering the same element of the experimental condition (ontology) c r , which have also been annotated as positive.…”

“…In the last years, several researchers have emphasized the use of multitasking (mtk) computational models, which are able to integrate different kinds of chemical and biological data, allowing the assessment of multiple biological activities against diverse biological systems (microorganisms, cell lines, etc.). [33][34][35][36] A peculiar detail of all these mtk-computational models is that they are based on the use of graphtheoretic invariants (topological descriptors), which allow the characterization of the molecular diversity and complexity at local and global levels. 37 On the other hand, topological descriptors are very useful for determining the relationships between the substructures/fragments and the biological effects.…”

Artificial intelligence and machine learning‐aided drug discovery in central nervous system diseases: State‐of‐the‐arts and future directions

“…To benefit from the data obtained from multiple assays, researchers aim to develop multitask QSAR models. Several groups constructed the multitask learning structures based on plain feed‐forward NN to avoid overfitting by learning multiple bioassays simultaneously 190–196 . Moreover, multitask QSAR models were also utilized for predicting the activity against multiple targets 197–199 …”