AI/ML in the Schrödinger Suite – Update on Machine Learning Applications in Drug Design
Featuring Dr. Thomas Steinbrecher, Senior Director at Schrödinger

The Schrödinger molecular design platform integrates solutions for physics-based modeling, data analytics, and collaboration together with AI/machine learning methods, to enable more rapid discovery of multi-parameter optimal new molecules. Multiple examples of enterprise-wide deployment of these technologies have been demonstrated to increase the success for e.g. preclinical drug discovery. Here we will present case studies of novel machine learning methods recently added to the Schrödinger Technology Platform.

Neural networks trained on DFT energy calculations have emerged as exciting alternatives to fixed-form potential energy functions such as molecular mechanics force fields in computational chemistry applications. We describe the development of the Schrödinger-ANI potential with near complete coverage of drug-like chemical space. The performance of Schrödinger-ANI in common computational drug discovery applications will be discussed.

We have previously demonstrated a large-scale exploration of chemical space in the hit-to-lead process by coupling reaction-based enumeration, active learning, and free energy calculations. Here, we augment that approach by adding goal-directed generative machine learning models, which results in a higher enrichment of potent ideas compared to large scale enumeration alone. We examine the utility of this combined approach by designing potent inhibitors of cyclin-dependent kinase 2 (CDK2) and demonstrate a coupled workflow that can provide improved enrichment in identifying < 10nM compounds over our previous method. The reported data suggest combining both reaction-based enumeration and generative machine learning for ideation enables the rapid discovery of novel chemical matter within a predefined potency and property space.

Structure-based Anti-Viral Design using Schrödinger’s Biologics Modeling Suite
Featuring Dr. Jianxin Duan, Schrödinger Fellow

Since prehistoric time, virus infection has been a threat to human health. Modulation of immune response is one of the successful approaches to prevent and treat infection. Vaccines provide sustained protective immunity by stimulating immune response of the host. Originally, vaccine design was centered around the isolate-inactivate-inject paradigm. In recent years, advances in structural biology and other technologies have introduced the era of rational vaccine design in which recombinant protein antigens are modified to achieve desired properties. Antiviral antibodies treat infections by rapidly removing virus particles and infected cells. Structure-based design of antibodies plays an important role in improving crucial biochemical properties such as affinity and aggregation. During this webinar, we show how Schrödinger's biologics modeling suite can be used in the engineering of protein antigens and antibodies to improve safety, efficacy, and developability of the therapeutic agents.

In Silico Screening of Plant Small Molecules to Bind ACE2 Specific RBD on Spike Glycoprotein of SARS-CoV-2 for Repurposing Against COVID-19
Featuring Dr. Hrishikesh Damle and Dr. Brarath B., Atrimed Biotech

Human coronavirus (2019-nCoV, also known as SARS-CoV-2) is causing a pandemic with significant morbidity and mortality. Although, no effective novel drugs targeting SARS-CoV-2 are available, drug repurposing is emerging as an effective drug discovery strategy from existing drugs. The strategy can be beneficial because of its shortened discovery time, reduced cost and already clinically proven safety for human use.  In this study, we present an in silico drug repurposing study implementing successful concepts of computer aided drug design (CADD) technology. The objective of the study is to repurpose known plant molecules to interfere the viral cellular entry via the spike glycoprotein (SARS-CoV-2 S), which mediates the virus-cell receptor interaction. SARS-CoV-2 S uses ACE2 to enter cells. In the present study, we screened 4,08,000 plant small molecules in our library to evaluate their interaction with SARS-CoV-2 S using molecular docking studies, out of which 15 lead molecules were shortlisted. Further, one molecule was selected based on their reported anti-viral activity, safety, availability, and affordability. Further, the ligand-receptor complex was subjected for Molecular Dynamics (MD) simulation. The MD simulation results indicate the identified molecule from plant origin may be repurposed against -SARS-CoV-2 and further studies are needed for validation.