Newer Service, Development Information
Together with the University of Tokyo (Faculty of Pharmaceutical Sciences), Osaka University (Faculty of Pharmaceutical Sciences), and Drug Discovery Initiative, company’s collaborative research in the field of Pharmaceutical drug development, which requires our AI and computational chemistry technologies has been conducted. By our proposed method, the University of Tokyo selected the experimental compounds from the Drug Discovery Initiative’s compound library. The experiment is carried out in Osaka University. Detailed information will be announced on our company’s website as soon as the paper is published.
Upon request from AIST (National Institute of Advanced Industrial Science and Technology), a serverless computational chemistry environment was established. when you drag and drop the input files to the browser, the trigger will be pulled on, then the cluster environment will be activated, and various molecular simulations will start to perform. With the end of calculations, the cluster will be automatically shutdown. Some other research institutions are also considering using it.
At the request of Amazon Web Service Japan to expand the use of the cloud in drug discovery research sites, we created an AI model (for pharmacology). This technology has been successfully introduced to one pharmaceutical company and is still being improved.
NEXT COMPUTER-AIDED DRUG DESIGN
MolGate is a web interface that automatically creates computation resources required in the use of screening or molecular simulation on the cloud. With AWS on the back end, Each researcher can prepare a cluster environment with several clicks without the assistance of an IT administrator. We will offer a customized system for your company at the time of delivery.Please refer to the following movie about the demonstration which independently expanded the system provided to AIST.
myPresto on MolGateNEXT COMPUTER-AIDED DRUG DESIGN Domestic drug discovery sites, especially the environment surrounding in silico, have undergone significant changes in the last 5 or 6 years. Focus was completed in April 2011, and the domestic fastest supercomputer 京 was completed in July 2012, which has dramatically improved the precision of molecular dynamics simulation. As a result, a new era of silico drug discovery began, and has produced many academic / industrial results so far.
The era of cloud has been coming and it has been at the same time stirring up a spree of reforming the world’s server/ IT infrastructure one after another.
Under these circumstances, overseas drug discovery research centers has been actively promoting the use of clouds while utilizing supercomputers and computer centers, so as to reduce the calculation cost and also shorten the calculation time.
Pfizer’s high performance computing (HPC) software and systems for worldwide research and development (WRD) support large-scale data analysis, research projects, clinical analytics, and modeling. Pfizer’s HPC services are used across the spectrum of WRD efforts, from the deep biological understanding of disease to the design of safe, efficacious, therapeutic agents .Dr. Michael Miller, Head of HPC for R&D at Pfizer explains why Pfizer initially considered using Amazon Web Services (AWS) to handle its peak computing needs:
In 2013, Novartis ran a project that involved virtually screening 10 million compounds against a common cancer target in less than a week. Calculations estimated that it would take 50,000 cores and roughly a $40 million investment if they wanted to run the experiment internally. The project ran across 10,600 Spot Instances (approximately 87,000 compute cores) and allowed Novartis to conduct 39 years of computational chemistry in 9 hours for a cost of $4,232. Out of the 10 million compounds screened, three were successfully identified.
BRISTOL-MYERS SQUIBB -Bristol-Myers Squibb (BMS) used AWS to build a secure, self-provisioning portal for hosting research so scientists can run clinical trial simulations on-demand while BMS is able to establish rules that keep computing costs low. Compute-intensive clinical trial simulations that previously took 60 hours are now finished in just 1.2 hours on the AWS Cloud. Running simulations 98% faster has led to more efficient and less costly clinical trials and better conditions for patients.
AI / Cloud
・創薬研究におけるクラウド活用の実際 - myPresto × AWS - (At the Enterprise Session of Amazon Web Services Japan)
CBI2020・myPresto, computer-aided drug development software（Co-author）.
BSJ2019・Prediction of toxicity through the chemical space generated by deep learning
・Auto Cell Image Classification System for Micronucleus Assay by Deep Learning （Co-author）.
CBI2018・Dr. Kazuyoshi Ikeda (Keio University) introduced our AI works for chemical spaces.
JSBI2018・A Genetic Approach to Deep Learning in Prediction of Molecular Properties of Agrichemicals.
PSSJ2017・Dr. Yoshifumi Fukunishi (Advanced Industrial Science and Technology (AIST)) introduced our work that used Deep Learning for LogP prediction of middle molecules.
BSJ2017・Dr. Yoshifumi Fukunishi (Advanced Industrial Science and Technology (AIST)) introduced our work that applied AI model for the prediction of MD simulations .
Funatsu Lab.・A Genetic Approach to Deep Learning in Prediction of Molecular Properties.
CBI2017・An Application of Deep Learning for Classifying Chemical Structures.
・Towards Desktop laboratory from MD, docking, and virtual screening to bio assay: Computation performance （Co-author）.
CBI2016・Scalability and performance of applications on cloud computing: virtual screenings and MD simulations of myPresto suite （Co-author）.
IIBMP2016中分子・膜系を扱うための機械学習と計算シミュレーション技術の開発：myPrestoの新機能（Co-author）
JBiC / 2019Prediction of Passive Membrane Permeability by SemiEmpirical Method Considering Viscous and Inertial Resistances and Different Rates of Conformational Change and Diffusion（Co-author）
Molecular Informatics / 2019創薬における機械学習と分子シミュレーション：生体での膜透過現象 (特集：機械学習と分子シミュレーション) Vol. 22，No.2, April 2019 (通巻86号) （Co-author）
MSSJ / 2019機械学習と分子シミュレーションによる 天然物・中分子の物性推算（Co-author）
AIST / 2019中分子用物性推算への機械学習と分子シミュレーションの融合手法（Co-author）
JBiC / 2018We read a poster.
JBiC / 2017データベース利用での分子シミュレーション : 半シミュレーション的学習法 (特集 創薬で活躍する分子シミュレーション)掲載誌 19(4)=80:2017.10 p.245-255 （Co-author）
ArticleCentralクラウドとGPGPU：大規模化・複雑化する 計算化学を安全に速く利用する （Co-author）
株式会社 エー・イー企画Miscellaneous Topics in Computer-Aided Drug Design: Synthetic Accessibility and GPU Computing, and Other Topics （Co-author）.
ArticleCentral・Advanced Industrial Science and Technology (AI/Cloud)
・Toray Industries, Inc. (AI)
・Keio University (AI/Cloud)
・University of Tokyo (AI)
・Osaka University (AI)
・Amazon Web Service Japan (AI/Cloud)
About BY-HEX LLP
Limited Liability Partnership is a corporate organization formed on the basis of a partnership agreement aimed at the project. For all partners, their responsibilities are limited. [Excerpt from www.wikipedia.com]
We believe that the answer is NO. It is not realistic for the busy researchers to do AI program as the cost to catch up the trend of the framework which changes fashion as often as every half year or even every one year and put the paper contents into practical are both far too high. Researchers are mathematically able to study and analyze theme setting and use other strategic methods to logically investigate correlated papers. However, by providing the learning environment (the cloud environment included) and implementing the program, we could succeed to reduce the cost of introduction and development to much lower point.Generally speaking, although the problem of the drug discovery field has become complicated and fragmented in recent years , a lot of researchers, not only in AI field but also in other various related field , are working together to make the situation better.
Due to the fierce competition between cloud service companies, the prices of essential computing resources for research institutions and development companies are on the decline. In the second half of 2016, you could borrow a machine equipped with high-performance GPU (K80), necessary for deep learning, with only $1.50 (even in Japan). With the advent of GPU compatible docking environment brought by the appearance of nvidia-docker, the difference between the cloud and the local computer was minimalized in early 2016. BY - HEX believes that efficient management of computing resources and coordinating a large amount of learning data is a key to AI development.
