Artificial intelligence in drug discovery

Artificial intelligence helps scientists in both steps of drug discovery, genesis of the targets as well as in design and screening aspects, say Kole P L, Girish Bende, Bhusari Sachin, Nagappa A N

The emergence of the human genome has opened new frontiers in the drug discovery process at molecular level as it ameliorate our understanding of genesis and progression of various diseases of self and acquired type. This has resulted in obtaining so-called ’molecular targets’ for drug design and discovery process. If such a molecular target is identified, the search for those molecules begins, which influence the target’s activity specifically and which are, therefore, considered to be the most potential and selective drugs against the disease.

Furthermore, the advancement of combinatorial chemistry, recombinant DNA technology and solid phase synthesis technologies has bestowed scientist with thousands of compounds, generating high demand of screening for biological activities, to be considered for further clinical studies by attrition of the compounds with less activity at early time. Screening of single chemical entity for single and multiple targets results in huge experimental data. Processing this myriad data obtained from such unit study demands, advanced computational tasks and logical decisions, which, in turn, entail huge manpower, consistent mental attention, and limitless time.

Artificial intelligence (AI) has come as a unique tool, which provides us with high processing speed with highest fidelity and accuracy, data storage and retrieval, overcoming the boundaries of the human limitations with multiple computations and logical decisions in a programmed approach. Moreover, AI helps scientists in both steps of drug discovery, genesis of the targets as well as in design and screening aspects.

Intelligent High-Throughput Screening (HTS) is a promising tool for drug discovery that has gained widespread popularity over decade, which synonyms for the fuel for the drug discovery machine. HTS is the process of assaying a large number of potential effectors of biological activity against targets (a biological event). The methods of HTS are applied to the screening of combinatorial chemistry, genomics, and proteomics, including protein and peptide libraries.

The goal of HTS helps expedite the drug discovery by screening large array of compounds often composed of hundreds of thousands of new chemical entities (NCE), at a rate that may exceed 20,000 compounds per week. Advancement of the robotics in screening, have enabled and catalysed the unprecedented increase in assay throughput with commencing new field as Ultra high-throughput screening (UHTS), which facilitates the testing of 100,000 compounds/day.

Due to the need to process thousands of assays per day, HTS has revolved around the combined world of multiple-well microplates and robotic processing. For a number of years, HTS assays have been run in the standard 96-well microplate (working volume of up to 250 L). The current research interest of most organisations is demanding beyond this format to higher-density, lower-volume formats (e.g., 384- and 1536- well microplates).

There are two primary advantages of these formats: increased throughput and lower volume, which translates into lower cost. At screening rates of 500,000 compounds/week, a cost of $1 per well is difficult for any company’s budget to support on a weekly basis.

HTS is only one step in the early drug discovery process. Other steps include compound library construction, secondary screening, and compound library optimisation through medicinal chemistry. Assay/target development is a rate limiting step in drug discovery process for many research agencies. Basic considerations in designing highthroughput screening assays for a drug discovery are:

• Homogeneous assays for highthroughput and ultrahighthroughput screening
• Microbe-based screening systems
• Molecular genetic screen design for agricultural and pharmaceutical product discovery
• Receptor screens for small molecular agonist and antagonist discovery
• Functional assay screens
• Enzyme screens
• Screening strategies for ion channel targets for various leads
• Highthroughput screening assays for detection of transcription
• Screening of combinatorial biology libraries for natural products discovery
• Higherthroughput screening assays with human hepatocytes for hepatotoxicity, metabolic stability
• Highthroughput screening for metabolism and drug-drug interactions
• Single molecular spectroscopy and miniaturised genomics/functional proteomics for identification of new targets
• Bioinformatics: Identification of novel targets and their characterisation
• Laboratory automation
• Robotics and automation
• Assay miniaturisation: Developing technologies and assay formats.

The basic approach for the ideal drug discovery is three-dimensional view of physicochemical properties, pharmacodynamic properties and pharmacokinetic properties. Throughput has to be increased for all the designed compounds for these three basic requirements. The highthroughput assays are now available for the fast screening of the compounds for their physicochemical properties like ionisation constants (pKa), partition coefficient (Ko/w), solubility, etc with the deep understanding basic chemical information of the molecules and the mathematical algorithms.

For highthroughput drug pharmacokinetic studies, various caco-2-cell based highthroughput absorption assays, highthroughput drug metabolism assays are available, thus reducing the time for complete profiling of the drug candidate. Highthroughput methods and assays are now available for all three dimensions of the requirements of the drug and its discovery.

Various software based intelligence systems for drug discovery are:

1) Assisted model building with energy refinement (AMBER): It is a set of molecular mechanical force fields for the simulation of biomolecules along with a package of molecular simulation programs.

2) Automated docking of the flexible ligands to macromolecules (AutoDock): It is a suite of automated docking tools. It is designed to predict how small molecules, such as substrates or drug candidates, bind to a receptor of known 3D structure. With additional support of X-ray crystallography, structure-based drug design, lead optimisation, virtual screening (HTS), combinatorial library design, protein-protein docking, chemical mechanism studies

3) Molecular Analysis Pro: It has a physical property estimation program, a 3-D chemical structure drawing program, a chemical data base creation program, a molecular graphics modeling tool, a reaction/mixture editor, a computer slide show maker, a batch structure printing program, an unsophisticated structural/reaction searching program.

4) MolSuite and MolSuite DB: It helps in molecular modeling with advanced graphics, physical property calculations, statistical analysis, and database development and management.

5) ChemSite3D molecular visualisation software and ChemSite Pro: These provide fast minimisation and displaying of any molecule, even crystals, in a 3D environment. Similar versions CS ChemOffice Pro, standard, ultra and CS Chem3D Pro, standard, ultra are available.

6) HyperChem: It is a molecular modeling environment equipped with 3D visualisation and animation with quantum chemical calculations, molecular mechanics, and dynamics.

The artificial intelligence and HTS, hand in gloves, continues to be competitive and dynamic. Both the systems have contributed to a great extent for reducing the time for the drug discovery and to obtain the successful drug candidate in a shorter period of time. The most important aspects of all the basic inventions lie in industrial perspective. Many of the research organisations and companies have come up with ready to use HTS technologies for accelerated drug screening. Reducing the total cost and time of the drug discovery, HTS is very well accepted by pharmaceutical companies and has been significantly driven by implementing technologies from vendor companies rather than through developments occurring within drug research companies.

The writers are with Pharmacy Group, Birla Institute of Technology and Science, Pilani, Rajasthan 333031.