Artificial intelligence: Machine learning approach for screening large database and drug discovery

Abstract

Recent research in drug discovery dealing with many faces difficulties, including development of new drugs during disease outbreak and drug resistance due to rapidly accumulating mutations. Virtual screening is the most widely used method in computer aided drug discovery. It has a prominent ability in screening drug targets from large molecular databases. Recently, a number of web servers have developed for quickly screening publicly accessible chemical databases. In a nutshell, deep learning algorithms and artificial neural networks have modernised the field. Several drug discovery processes have used machine learning and deep learning algorithms, including peptide synthesis, structure-based virtual screening, ligand-based virtual screening, toxicity prediction, drug monitoring and release, pharmacophore modelling, quantitative structure-activity relationship, drug repositioning, polypharmacology, and physiochemical activity. Although there are presently a wide variety of data-driven AI/ML tools available, the majority of these tools have, up to this point, been developed in the context of non-communicable diseases like cancer, and a number of obstacles have prevented the translation of these tools to the discovery of treatments against infectious diseases. In this review various aspects of AI and ML in virtual screening of large databases were discussed. Here, with an emphasis on antivirals as well as other disease, offers a perspective on the advantages, drawbacks, and hazards of AI/ML techniques in the search for innovative treatments.

Introduction

The vast research in the biological field has generated huge data. However, experts have questioned if biological data transmission and transfer are done effectively in order to derive practical knowledge (Sanal et al., 2019). In pharmaceutical research data is available in various forms such as research outcomes, clinical data, and ethnic population-wise data. The average cost of a new drug discovery and development process costs from $1 to $2 billion and takes about 15 years to release it in the market. When research problems are taken into consideration, the data can be used to develop new drug treatments that can be accurate and cost effective (Sousa et al., 2019).

The data suggest that almost 90% of drug candidates fail in the clinical phase, this makes huge production. The chemical database is so vast any technique that is quick, effective, and relatively inexpensive can be utilised to analyse these molecules. Many techniques are involved in computer aided drug discovery (CADD) such as virtual screening (VS), pharmacophore mapping, docking studies (Vamathevan et al., 2019a, Vamathevan et al., 2019b). VS is a computer technique for searching huge libraries of tiny molecules (ligands) for lead compounds that can bind to a molecular target. The Traditional screening method is a very tedious and laborious task so, in recent years artificial intelligence (AI) has drawn attention due its robust application (Paul et al., 2021).

To find out new suitable molecules the researchers are involved in developing new innovative methods and algorithms. Significantly artificial intelligence (AI), deep learning (DL), machine learning (ML), and computational chemistry are new approaches involved towards drug discovery. Such methods can be used alone or in combination to create novel technique that incorporate a diverse set of efficient algorithms that improve predictions (Mak and Pichika, 2019).

The application of AI-based computational techniques in pharmaceutics and health science are most stable and most effective. AI works by mimicking human performance in machine learning models (Chan et al., 2019). Computational models, machine learning algorithms play an important role in developing a new drug by predicting its various activities. The AI can help screen and identify the complex chemical structure, its properties and activity process in drug identification process. It has application in every field of pharmaceutics (Vanommeslaeghe et al., 2015). The integration of AI and ML is more effective and generates accurate predictions for newly designed drugs. In recent years, various types of study provide deep atomic insights which help to find out the sources of diseases, functions and other information (Ahuja, 2019).

Natural products are constantly being investigated in the development of new bioactive molecules with commercial applications, catching the interest of scientific research efforts due to their pharmacophore-like structures, pharmacokinetic features, and distinct chemical space. The bioactive compounds are more valuable to develop new drugs because of their physic-chemical diversity & less toxicity properties (A Bryce and H Hillier, 2014). From ancient times many bioactive compounds like pilocarpine, quinine, morphine, and artemisinin. They are from synthetic compounds in various aspects of the chiral centre, high percentage of oxygen, and their chemical space is highly diverse, containing different structural scaffolds, when compared with synthetic compound libraries. The bioactive compounds now became a new target for synthesis & development drugs because of their unique characteristics (Davenport and Kalakota, 2019). The conventional method of new bioactive compounds involves various steps starting from collection of material, authentication. Of sample, extraction, isolation, characterization of isolated bioactive compounds and finally biological assay to predict its activity. Further steps include bioactive/lead optimization by chemical synthesis, structural improvement through pharmaco-dynamic and pharmacokinetic properties and to increase their biological activities. While CADD is efficient, economic, less time consuming as compared to the methods used only in-vitro and in-vivo assays (Aguiar-Pulido et al., 2013).

The drug designing by CADD has two methods-structure-based drug design (SBDD) and ligand-based drug design (LBDD), both methods based on algorithms, scoring functions, and force fields for ranking. Nowadays many computer programming and software are developed that run with various algorithms to interpret the results of both methods using predefined scoring functions. But the prediction of exact energy levels and force field is a difficult task for screening the possible drug molecules. This can be solved by quantum physics which can increase the efficiency of prediction of new drug discovery and reduce the error (Zador et al., 2021). The advantage of CADD as well as molecular docking is it has ability to screen the complete small molecule database in fast and can show realistic interactions between hit molecules and macromolecules. The macromolecules are polymers of amino acids or nucleic acids which can be predicted by algorithms that can adjust the atoms in macromolecules. There are several freely available online tolls, programmes and software which can predict the properties of lead molecules. Recently different classes of natural product such as alkaloids, flavonoids, steroids, and terpenoids have been proven new drug candidates (Rayan et al., 2017). Apart from these phytochemical groups metabolomics products, bacteria, cnidaria, insects are also used in drug discovery. Currently the AI in silico analyses, artificial intelligence and cheminformatics approach have proved robust to selected interested candidates from large database and analysing their bioactivity, pharmacodynamics, and their pharmacokinetic properties (Newman and Cragg, 2016).

In this review, we discuss current computational methods that use artificial intelligence, machine learning, and cheminformatics, to screen the large database that have been developed in the last years.