Video Feature
Vol.13, September 2018
Yutaka Akiyama
Professor, Department of Computer Science
Head, Middle Molecule IT-based Drug Discovery Laboratory (MIDL)
Tokyo Institute of Technology
Why is Tokyo Tech pursuing research on “middle molecule drug discovery”?
Tokyo Institute of Technology (Tokyo Tech) is internationally recognized for its contributions to research in biology and chemistry. But some people may be surprised to hear that we have started research on drug discovery. Our interest in this area of research was triggered by the recent major changes in drug discovery research.
Drug discovery using "macromolecules" called antibodies is currently attracting attention for cancer treatment and so on. Macromolecules strongly bind only to specific targets, and although this approach is highly effective, the manufacturing process is very expensive, which is a social problem.
On the other hand, drug discovery using "middle molecules" such as peptides and nucleic acids, enables the synthesize of low cost drugs by combining predetermined parts. Also middle molecules are not as large as antibodies, so they can go through cell membranes and target the insides of cells. It is said that middle molecule drugs take the best features of traditional small molecular medicine, and antibody medicine. However, there are still many unresolved issues for middle molecule drug discovery to become mainstream. We expect that chemical synthesis technology and supercomputer analysis technology developed by Tokyo Institute of Technology will greatly contribute to overcoming these concerns.
What supercomputer can do?
Since the structure of the middle molecules is much more complicated than conventional low molecular weight drugs, the time and effort required for computer simulations to determine their properties becomes orders of magnitude greater. In addition, since it is difficult to use statistical prediction methods that were developed for small molecules, we are completely reconstructing prediction models by machine learning. In order to realize such calculations in short periods of time, we are utilizing TSUBAME supercomputer possessed by Tokyo Institute of Technology.
Details of technology development
1. Prediction of internal sustainability
In 2017, we established MIDL, Middle Molecule IT-based Drug Discovery Laboratory. Experts in chemistry, biology, pharmaceutical science, information science, and social science at Tokyo Tech got together with the goal of resolving issues related to middle molecule drug discovery, with emphasis on utilizing information technology (IT). In addition to conducting research at Tokyo Tech's Ookayama Campus and Suzukakedai Campus, we also have a laboratory here at Tonomachi KING SKYFRONT, where we focus on studying drug discovery using peptides.
The first pillar of our research is to use computers to predict the sustainability of the effects of drugs after injection of peptides into the body. For example, if peptides are completely decomposed within 30 minutes, then it is not practical for drug applications because it is necessary to have appropriate levels of sustainability.
In scientific terms, we are developing technology to propose which parts can be improved by quickly determining something referred to as the "PPB value" by using only calculations and without experiments. In this theme, we mainly utilize supercomputer-based machine learning methods.
Details of technology development
2. Prediction of cell membrane permeability
The second pillar of our research is predicting whether peptides can penetrate cell membranes and reach intracellular targets. We are developing technology that can suggest which part of the drug should be improved. This theme mainly utilizes molecular dynamics simulations using our supercomputer. Membrane permeation of middle molecules takes a few milliseconds, which is a mere moment from the human sense, but it is a tremendously long process in the world of molecular world time. In order to enable long time simulations that was impossible in the past, we have implemented a new calculation method using the TSUBAME supercomputer.
What was the attraction to research and development of medicines using computers?
During my younger days, I was a researcher in an entirely different field called neural networks, but I was very interested in biology and drug discovery. I was fortunate that I have always been working closely to cutting edge computer development projects. The current computer is one billion times faster than ones during my younger days. But I used to think, how should you use computers as they become faster and faster? I discussed this with people from many different research fields. Currently, we make full use of the power of supercomputers for drug discovery and genome analysis.
Plans and schedule for commercialization
Regarding peptide drug discovery, we have been collaborating for several years with PeptiDream Inc. that is also located at Tonomachi KING SKYFRONT. We are also considering working with IT companies in the Kawasaki City area. Of the projects mentioned earlier, we are aiming to commercialize the prediction of internal sustainability of peptides in 2019. Regarding membrane permeability prediction, although data is still insufficient, we are planning to commercialize it one or two years later than sustainability prediction.
Expectations of Tonomachi KING SKYFRONT
Institutes related to our research and development are gathered at KING SKYFRONT, so discussion progresses quickly. Keio University is also in the RGB2 building as the Tokyo Tech MIDL, and we are using the computer server room jointly and also collaborating on a research project. In addition to PeptiDream Inc. mentioned earlier, we are currently conducting joint research with the National Institute of Health Sciences. I am very much looking forward to more and more excellent researchers gathering at Tonomachi KING SKYFRONT.
