AIDS is one of the multifaceted diseases, and this underlying complexity hampers its complete cure. In addition to that, HIV-infected patients have an increased risk of comorbidities, such as infectious diseases as well as noncommunicable disorders. Multitarget drug therapy is considered a better treatment option for HIV and infections as compared to targeted drug therapies and helps to overcome drug resistance, reduction in drug dosage and toxicity. In such a scenario, machine learning approaches can be used to alleviate the process of drugs discovery and development.<br/>We attempt to address the existing challenges by developing a computational approach to predict multitarget agents with anti-HIV properties effectively and as a prediction platform for the exploration of the Synthetically Accessible Virtual Inventory (SAVI) library. SAVI includes about 283 million molecules annotated with a proposed one-step synthetic route from commercially available starting materials. The SAVI database is well suited for ligand-based methods of virtual screening to select the most promising hits for experimental testing.<br/>Among the various therapeutic targets for HIV treatment, protease, reverse transcriptase, integrase, fusion, viral protein R, tumor susceptibility gene 101, viral infectivity factor, Rev, Tat, GP120, toll-likes receptors, P24 capsid protein, HIV-LTR and other are the primary focus. Besides, it should be noted that there are no less than 100 different HIV comorbidities diseases. We have developed freely available web service to select new potential anti-HIV agents based on selected targets from SAVI. We applied the computer program PASS (Prediction of Activity Spectra for Substances) and similarity estimates based on MNA (Multilevel Neighborhoods of Atoms) and QNA (Quantitative Neighborhoods of Atoms) descriptors. PASS has been developed and applied for drug design & discovery during the past 25 years and has been validated in more than 500 independent papers reporting on the application of PASS predictions to the discovery of new pharmaceutical agents.<br/>The combined multitarget approach of machine learning and similarity assessment allows estimating the anti-HIV potential of new compounds. This approach led to the selection of dozens of molecules recommended for synthesis and testing for antiretroviral activity, we have identified a few promising hits against the identified targets which may give new drugs to combat HIV infection after wet lab validation.

Freely available online resource for prediction of novel multitarget anti-HIV agents

2019 Spring National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.