Development of the next-generation simulation drug discovery method by integrating experiments and computation

Science related to drug discovery begins with molecular biology and organic synthetic chemistry from the upstream of the process, not only applied science in structural science such as structural biology, pharmacology, pharmaceutical science, but also drug candidates are clinically commercial It is a comprehensive science that requires knowledge of social science such as medical statistics, regulatory science, medical economics, and so on until it becomes usable, and like a case of biopharmaceuticals and molecular biology, a new science Innovation has occurred every time field knowledge is applied to drug discovery, and blockbusters (medicines that will raise sales of 100 billion yen a year) have been created by applying these findings.

However, because interdisciplinary research that can inherently cause innovation by using existing historical academic fields and new academic fields is a drug discovery science that encompasses many specialized fields, it is difficult to fuse and communicate between academic fields It has become to.

Personally, I think that its characteristic case is a barrier between in silco drug discovery (calculation, Dry) and compound screening, field optimization and field (experiment, Wet) done. In the early 2000s purchasing over 1000 kinds of compounds that cost more than a few tens of thousands of yen by using the results of calculations by computational science experts installing large computers that are nearly 10 million yen even if they are cheap There are times when SBDD can not be implemented in the research budget of public research institutions such as universities very much, such as evaluation of bioactivity and finally obtaining several hit compounds. It seems that such enormous cost and expertise barriers impeded communication between Wet and Dry.

However, in recent years, if you have knowledge of programming more than a certain level, nearly all calculations necessary for in silico screening are possible with myPresto developed by Fukunishi of AIST and provided free of charge. In addition, we use a cloud computing service like the rental server of Hokkaido University Information Infrastructure Center or Amazon Web Services which we are also using to compare large-scale computers to shorten the calculation time, compared to before Calculation can be carried out at low cost. In addition, if the National Compound Library provided by the University of Tokyo Drug Dispensing Organization etc. is utilized, it is necessary to contract and report separately with the management agency, but it is also possible to create an environment where more than 1000 compounds can be acquired with only tens of thousands of yen It has become to. Unfortunately, although the evaluation cost after that is almost the same as in the previous situation, it seems that the overall cost is enough to be practicable in one laboratory of public research institutes such as universities. In addition, it is becoming impossible for an individual researcher to become familiar with both findings and conduct SBDD research.

Docking simulationにおける並列計算概略図

In order to appropriately set the calculation parameters that must be set when conducting in silico research using a parallel server, although the related software has become relatively user-friendly, it is necessary to know basic programming knowledge Deep experience and knowledge on drug discovery research are necessary. We believe that fostering such human resources and creating success stories in the academia is the only way to help eliminate such Wet and Dry barriers and it is important to identify infectious disease targets with particularly high medical needs such as viruses and Pseudomonas aeruginosa , We are working on drug discovery research using SBDD.