Korea University

KOREA UNIVERSITY


HOME

now page

Research

게시판 -- 목록(갤러리)
Professor Kang Jae-woo's research team and El Mito (LMITO) Thera...
  • 글쓴이 : Communications Team
  • 조회 : 169
  • 일 자 : 2020-05-15


Professor Kang Jae-woo's research team and El Mito (LMITO) Therapeutics have opened a path to shorten the development period of new drugs using AI

 

After confirming the target, the work that took 1 ~ 2 years to derive the lead compounds was dramatically shortened to 10 weeks.

Artificial intelligence technology was used to derive lead compounds for treatment 

of severe neurological diseases

The inhibitory ability against disease target proteins was verified to the cell level.


▲ From the left, Kang Jae-woo (Professor of Computer Science, Korea University), 

Lee Eun-joo (Executive Director at El Mito Therapeutics), 

Hong Yong-rae(Vice President, El Mito Therapeutics), and Lee Hui-sung(CEO, EL Mito Therapeutics)

 

Professor Kang's team at the Department of Computer Science at the College of Informatics announced that AI technology has been used to derive a lead compound for treating severe neurological diseases.

 

The team and El Mito Therapeutics, an innovative new drug development bio-venture company, succeeded in deriving lead compounds for the treatment of severe neurological diseases within 10 weeks upon starting a study after signing a joint research agreement in February this year.

 

Among the top 50 drugs derived from the AI new drug development platform built by the research team, El Mito Therapeutics selected 23 drugs based on the activity and property prediction results of the 3D structures of disease target proteins. The team of Hanyang University’s Professor Yang Chul-su conducted cell activity experiments on the first 11 drugs. As a result, activity was confirmed in all 11 drugs, and especially, 2 of them showed high activity at the nanomolar level.

 

Usually, it takes about 1 ~ 2 years to derive a lead compound after the target is identified. The results of this study are significant because it represents a successful case in which the time to derive was shortened to 10 weeks by utilizing AI. Even more encouraging, the activity was confirmed at the cellular level for all 11 drugs tested. This was possible because the team’s AI platform learned to predict drug effects at the intracellular transcript level, away from traditional target-driven new drug development. In other words, it is designed to find a drug that induces a diseased intracellular gene expression pattern into a normal intracellular gene expression pattern, so that it is possible to directly derive a drug having activity at the cell level, beyond the activity at the protein level. During the 10-week period, the AI took 2 weeks to derive the drug, and it took the remaining 8 weeks to prepare the drug and confirm the inhibitory activity results at the cellular level.

 

The AI new drug development platform is built on models that have placed high at various international biomedical AI Challenge competitions over the years.

 

In 2016, the team participated for the first time in the AstraZeneca Sanger Drug Synergy Prediction DREAM Challenge and placed second, outplacing Stanford University and MIT. The following year, the team received an award for being the best performers at the NCI-CPTAC DREAM Proteogenomics Challenge hosted by the National Cancer Institute (NCI) Clinical Proteomic Tumor Analysis Consortium (CPTAC), outplacing UCLA, MD Anderson Cancer Center, etc. to become the first Korean team to win the Dream Challenge. In 2018, the team also won the Multi-targeting Drug DREAM Challenge hosted by the Icahn School of Medicine at Mount Sinai Medical School in the U.S., beating out the multi-national pharmaceutical company, Janssen Pharmaceutica. In 2019, the team became co-winners with the University of Illinois-Chung Hua University Consortium and the University of North Carolina at the AI-based Drug Development DREAM Challenge by beating out the representative government-funded research institutes of the U.S. and Europe, the National Institutes of Health (NIH) and the European Molecular Biology Laboratory (EMBL). In doing so, the team achieved the notable accomplishment of winning the DREAM Challenge in three consecutive years. Started in 2007 and with this year marking its 14th year, the DREAM Challenge, hosted by IBM and Sage Bionetworks, is the most distinguished international biomedical AI competition. In addition to the Dream Challenge, Professor Kang’s team participated for the first time in the BioASQ Task 7B-Phase B Challenge held in October of last year, sponsored by Google, the National Institutes of Health (NIH), and the European Union. The team finished in first place in all 5 rounds of evaluation beating out runner-up Google by a wide margin and received recognition for its unrivaled skills.

 

Professor Kang said, “Our AI new drug development platform, which is a platform that combines the element technologies recognized through international competitions for the past 4 years, has been verified to some extent through this success story, and in the future, it will be possible to develop successful new drugs for severe neurological diseases by synergistically combining the candidate substance optimizing technology accumulated by El Mito Therapeutics.”

Research 게시판 리스트