Foreword

The development and applications of in silico technologies has been active in the field of drug discovery over the past decades. Recent developments in deep learning have led to further advances of artificial intelligence (AI), and in silico technology is attracting more attention than ever before. In silico technologies in the field of pharmaceutical sciences are diverse ranging from quantitative structure activity relationship (QSAR), statistical machine learning, cheminformatics, molecular dynamics simulations, quantum chemistry, etc., each of them consisting of different fundamental technologies. Hence, it is not easy to understand these technologies in an integrated manner and to select the appropriate technology for a particular problem. Therefore, I have been planning this Current Topics for about two years to introduce recent applications of in silico technologies in drug discovery.

All of the first authors of the invited reviews comprising this Current Topics are members of the Division of Structure–Activity Studies (DSAS) of the Pharmaceutical Society of Japan. However, this Current Topics was organized independently of the division. The members of DSAS consist of researchers interested in a wide range of in silico technologies, covering all aspects of pharmacology, physics, medicinal chemistry, pharmacology, pharmacokinetics and toxicology, as well as researchers who are mainly involved in wet experiments. Readers interested in the activities of the DSAS are encouraged to attend events organized by the division.

The Current Topics in this issue contains four invited reviews describing in silico technologies and artificial intelligence for academic drug discovery. The first invited review, entitled “Compounds Designs of CK2α Inhibitors Derived from Virtual Screening Hit Compounds by Computational Chemistry with Crystallography” has been written by Dr. Nakamura et al. Novel CK2α inhibitors with diverse scaffolds were identified through virtual screening utilizing a unique approach called SDO-VS: virtual screening by shape-mimetics for water molecule dynamics in ligand binding site. From two hit compounds with different scaffold, X-ray crystallographic analysis of the complexes and molecular design using in silico techniques led to the efficient enhancement of inhibitory activities and physicochemical properties in both series.

The second invited review, entitled “Computer-Aided Drug Design Using the Fragment Molecular Orbital Method: Current Status and Future Applications for SBDD” has been written by Dr. Takaya. A structure-based drug design (SBDD) approach using quantum chemistry calculations was studied to evaluate protein–ligand interactions. Computer-aided drug design has become an essential tool for accelerating and optimizing the drug development process, and the Fragment Molecular Orbital (FMO) method plays a role as one of the effective tools for interaction analysis by introducing quantifiability to interaction analysis. In the review, the authors outline the recent progress in SBDD, including a case study that FMO interaction analysis for an experimental structure of hematopoietic prostaglandin D synthase (H-PGDS) with a high-potent inhibitor, and mention the potential of machine learning to utilize large-scale computational datasets for drug development.

The third invited review, entitled “Interaction Analysis by Fragment Molecular Orbital Method for Drug Discovery Research” has been written by Dr. Kawashita. The FMO method was used to analyze protein–protein interactions in order to find inhibitors of protein–protein interactions and to elucidate the pathogenicity associated with protein–protein interactions. Furthermore, they have applied this method to the calculation of antigen–antibody interactions and integrated it with molecular dynamics calculations to search for inhibitors that also take into account dynamic conformational changes. These methods are expected to be applied not only to small and large molecule inhibitor, but also to mid-molecule inhibitor and molecular design using non-natural amino acids.

The fourth invited review, entitled “Development of Drug Discovery Platforms Using Artificial Intelligence and Cheminformatics” has been written by Dr. Kawai et al. An evolutionary approach to designing new compounds by combining molecular fragments has been proposed and its performance is being evaluated. Machine learning models have also been developed for virtual screening and large-scale activity prediction, with the aim of predicting side effects. An attempt to use AI to assess the binding mode of a drug molecule to its target protein is described. Applications of informatics and AI in drug discovery are also presented.

I believe that this Current Topics will be informative for readers to further understanding of in silico drug discovery. Finally, I thank the Editorial Committee of the Pharmaceutical Society of Japan for accepting my proposal of this Current Topics. In addition, I thank all the authors for their significant contributions to this special issue of Chem. Pharm. Bull.