Current Trends in Computational Modeling for Drug Discovery

Proteins are the important biological macromolecules that are targeted by most of the existing drugs. SBDD play a critical role in design of drug-like, novel, potent, and safe modulators. It is a joint effort from structural biologists and computational scientists, which considers various limitations of the techniques and suitably guides drug designers. Identifying a novel, potent, and safe drug-like molecule is a long challenging path, and throughout this discovery journey, SBDD provides crucial guiding light at different stages. SBDD involves the use of structural data of target proteins to identify suitable ligand candidates that might bind the protein of interest and modulate its functions, resulting in therapeutic benefit. In this chapter, we provide an overview of computational SBDD workflow, and the various challenges associated with it. We also discuss strategies that could be adopted to tackle the challenges by making the best use of available information.

HDAC6 and HDAC10 are class IIb HDAC isoenzymes. They have unique structural and physiological functions. They are key regulators of different physiological and pathological disease conditions. HDAC6 and HDAC10 are involved in different signaling pathways associated with several neurological disorders, various cancers at early as well as advanced stages, rare diseases, immunological conditions, etc. Thus, targeting these two enzymes has been found to be effective for various therapeutic purposes in recent years. More work is still needed to pinpoint the selectivity as well as potency of class IIb HDAC inhibitors (HDACi) for their clinical development. The present chapter deals with the structural biology of class IIb HDACs and discusses how in silico studies including the virtual screening approaches have been implemented to design HDAC6 and HDAC10 inhibitors. In addition, the interactions of class IIb HDACs with their inhibitors are also highlighted extensively to get a detail insight. This chapter offers understanding for designing newer class IIb HDAC inhibitors in future.

Researchers are striving hard for the last two decades to discover new therapeutically effective molecules for the treatment of Alzheimer’s disease (AD). Unfortunately, the exact etiology of Alzheimer’s disease is not yet fully known, which is proving to be the main hurdle for the discovery of drugs to treat the disease. Several factors are involved in the etiology of AD, like oxidative stress, low levels of acetylcholine (ACh), β-amyloid aggregation, and tau protein phosphorylation in the brain. But unfortunately, no single drug has proved clinically effective till today to prevent or stop progression of the disease. The existing drugs simply improve the worsening clinical symptoms of AD and help in delaying the process of progression of the disease to a fully blown state. Currently, available drugs in the market for the treatment of AD include donepezil, galantamine and rivastigmine, the three acetylcholinesterase inhibitors (AChEIs), and memantine an N-methyl-D-aspartate receptor (NMDAR) antagonist. These drugs are used mainly to alleviate mild cognitive impairment (MCI) providing temporary relief from the symptoms. This chapter discusses about the application of various computational tools for compounds containing different heterocyclic moieties like quinoline, pyridine, pyrimidine, coumarine, chromane, indole, etc., which could serve as potential leads to design potent novel anti-Alzheimer’s agents.

As a result of the damage that viruses have done over time, humans have developed a variety of defenses against viral illnesses, such as vaccines and antiviral drugs for treatment. Since the 1950s, new viral illnesses including AIDS, Hepatitis, and coronavirus infections like SARS, MERS, and COVID-19 have periodically emerged, posing a challenge to the development of antiviral drugs. The creation of computer models is an interactive, iterative process that blends empirical datasets with known facts and assumptions (knowledge-driven or data-driven approach). In order to allow system simulation, the generated models should ideally offer reusability, composability, and interoperability. We surmise that the development of computational and mathematical frameworks will not only assist the development of newer antivirals, but simulating viral infections will also help in incorporating progressive immunosenescence and finding host genetic factors to expand the knowledge of infectious disease to an unprecedented level of detail. In addition to the fundamental molecular aspects of viral infection, this chapter emphasizes the fundamentals of computer modeling and discusses the relationship between in silico experiments and viral infections.

With over 36 million people currently living with HIV, HIV/AIDS continues to have devastating effects on human health worldwide. Viral resistance to anti-HIV drugs remains a major cause of concern, necessitating a regimen of highly active antiretroviral therapy (HAART), which consists of a combination of multiple drugs for long-term clinical benefit. Clearly, the rapid development of novel molecules that can help change the present regimen to new drug combinations is critical for tackling the resistance problem. In this regard, computational methods have emerged as a valuable tool in HIV research, contributing greatly to our understanding of HIV biology and aiding in the design of potent anti-HIV compounds. This chapter gives an overview of the various computational strategies reported in the discovery of drugs for the treatment of HIV. A comprehensive overview of several structure-based and ligand-based computational methods is presented first; this is followed by some notable applications of these methods in the discovery of novel anti-HIV compounds. Finally, we discuss the emergence of powerful machine learning algorithms which have proven useful both in the design of new compounds and in the development of theoretical models that can predict resistance to antiretroviral therapy.

Severe fever with thrombocytopenia syndrome virus (SFTSV) also known as Dabie bandavirus of the family Phenuiviridae is a negative-strand RNA virus and a tick-borne virus. Replication of SFTSV into systemic circulation and occurrence of viremia cause cytokine storm and T-cell overstimulation. The event of viremia-induced thrombocytopenia causes reduced platelet count and splenic macrophages, followed by endothelial damages and compromised immune system that cause multi-organ damages. Limited options for specific anti-SFTSV drugs pose significant challenges associated with clinical management of SFTSV infection. This book chapter chiefly emphasizes upon the genetic diversity, geographical distribution, pathogenesis associated with various clinical aspects like symptoms, diagnosis, and available clinical management options. In addition, current research linked with anti-SFTSV drug development is comprehensively portrayed in this review.

Toxicological aspects represent a fundamental step in the process of drug design and discovery. There are multiple platforms available, and recently freely available tools provided results comparable with those obtained from the commercial ones. We will present examples of models for the different endpoints which can be used. In addition, the future perspectives are to take into account in an earlier stage the adverse effects, in order to simplify the long process of drug design and discovery, and to optimize the selection of preferable features present in a new pharmaceutical. In this new vision, a more holistic approach can apply multiple methodologies and not only the screening of the adverse effects.

In silico approaches for activity/toxicity predictions have gained attention recently, and these are accepted by various regulations like EU-REACH. Aspects like reproducibility, less ethical complications, no animal use and reduced time are some of the reasons why researchers nowadays are shifting toward the in silico approaches for prediction. Quantitative Structure–Activity Relationship (QSAR) is one of the most commonly used in silico approaches for the prediction of response, but the only drawback is that since it involves model-derived predictions, it is prone to erroneous results when the number of training data points is insufficient. In recent times, similarity-based algorithms like Read-Across are being adopted by researchers with the aim of data gap filling. The Read-Across approach does not involve model-derived predictions, rather it involves similarity-based predictions and thus can efficiently be used for data gap filling. The authors at the DTC Laboratory have developed a Java-based Read-Across tool (https://​sites.​google.​com/​jadavpuruniversi​ty.​in/​dtc-lab-software/​home) which utilizes three different similarity-based approaches (Euclidean Distance-based, Gaussian Kernel Similarity-based and Laplacian Kernel Similarity-based) for the prediction of responses of the query compounds along with the external validation metrics and the overall error measures. Moreover, the computation of certain compound-specific similarity and error-based metrics enables the user to identify the uncertainty in the Read-Across-based predictions, especially when the observed response values of the query compounds are unreported. The idea of clubbing the QSAR methodology and the Read-Across approach together has given rise to a novel chemometric prediction approach termed as Read-Across Structure–Activity Relationship (RASAR). The authors at the DTC Laboratory are the pioneers in reporting the quantitative predictions using the RASAR approach (q-RASAR). A Java-based RASAR descriptor calculator tool has also been developed which calculates the similarity and error-based descriptors based on the similarity-based approach selected by the user. The authors feel that these tools have a lot of potential in bridging data gaps and may prove to be very much essential for the predictions of various property/activity/toxicity endpoints in the future.

Computational drug design and discovery have taken center stage attention during the time of COVID-19. The science community acknowledges the importance of ligand-based drug design (LBDD) and structure-based drug design (SBDD) to nullify the problem associated with a typical drug discovery process. In the modern era, a complement between experimental, theoretical, and computational approaches can make the drug discovery process rational, economical, and fast. Undoubtedly, computational power has increased manifold compared to the last few decades, making it possible to run many unthinkable calculations that cannot be imagined a few years ago. Along with the computational power, resources like open-access and commercial organic chemicals, phytochemicals, approved, experimental and investigational drugs, peptides, and metabolomic databases have increased enormously. Compared to designing a new drug, utilizing existing chemical and drug databases for virtual screening makes the process faster as the database chemicals are already synthesized (in most cases) and characterized. Even in a few instances, absorption, distribution, metabolism, excretion, and toxicity (ADMET) profiles are checked along with data for preclinical and clinical trials (primarily for investigational and/or in the process of approval drugs). A drug database is also a powerful resource for drug repurposing, where an old, approved drug for a specific disease can be used to treat another common/new/rare disease. The idea is increasingly becoming an attractive proposition as it comprises the use of already evaluated de-risked compounds which help lower the new drug development costs in a shorter time. Therefore, drug databases have an immense role to play as a repository of potential drugs for any common to a rare disease in the process of CADD and for the experimental scientists.