• Ұ/ҽ News&intro
  •  Faculty
  •  Events
  • к Undergraduate
  • п Graduate
  •  Career
  •  Admin
  •  Reservation
  •  Donation


  • Kim, Woo Youn   Associate Professor
  • Intelligent Chemistry
  • Ph.D, 2009, POSTECH
  • WEBPAGE : http://wooyoun.kaist.ac.kr
  • E-MAIL : wooyoun@kaist.ac.kr
  • Tel/Office : 042-350-2815 (Office), 2855 (Lab) / 3102(E6-4) (Room)

Contact information

Tel: (office) +82-42-350-2815, (lab) +82-42-350-2855
Location: (office) Room 3102 (Bldg. E6-4), (lab) Room 3119, 3120 (Bldg. E6-4)
Fax: +82-42-350-2810

Education

03/2004-02/2009 POSTECH, Chemistry, Ph.D
03/1997-02/2004 POSTECH, Chemistry and Physics, B.S

Professional Experiences

-10/2009-12/2010, MPI-Halle, Postdoctoral Fellow
-03/2009-09/2009, POSTECH, Postdoctoral Fellow
-10/2008-01/2009, Freie University Berlin, Visiting Scientist

Awards

-Best Lecture Award, KAIST, 2021
-Young Physical Chemist Award, Physical Chemistry Division of KCS, 2020

RESEARCH AREA


Brief Introduction

 

The ultimate goal of chemistry is to design molecules with desired properties and synthesize them on demand. This can be achieved only by being able to freely manipulate matter at the atomic scale. Sounds unrealistic for now, but the dream will be come true by continuing efforts toward the ultimate goal. To this end, we apply digital technology to chemistry. Traditional computational chemistry allows us to simulate chemical phenomena on a computer. These tools are based on physical principles such as classical mechanics, quantum mechanics, and statistical thermodynamics. However, they have been regarded as supplementary means of explaining experimental observations and validating intuitive ideas, because only human intelligence can design molecules and propose experiments for synthesis. Thus, intelligence is a key source of scientific creativity.

Our group challenges integrating artificial intelligence into computational chemistry. AI is transforming industry, science, and even daily life. Chemistry is not an exception. AI as a powerful data-driven approach can be the best complementary of the physics-based approach. We have already developed deep learning models for smart molecular design, especially focused on drug discovery. We expand this approach to designing materials such as OLED. We are also developing digital tools to automatically predict reaction mechanisms and synthetic pathways. Moreover, cloud computing gives unlimited power to the computational approach. In such a way, modern digital technology is changing chemistry. This is just the beginning of intelligent chemistry.

Research topics

Reaction discovery
Chemical reactions are complex multiscale phenomena. Quantum mechanics dictates microscopic changes, while kinetics and thermodynamics govern macroscopic changes. This high complexity prohibits prediction of chemical reactions via direct simulations. Here, we develop intelligent simulation tools powered by AI and use them to predict chemical reactions

Drug Discovery
Drugs are the most valuable organic compounds to cure diseases. However, developing a single approved drug is a long risky process; it takes 10~15 years and 2.6 B$ on average. AI can greatly accelerate this process through a very efficient data-driven approach. Here, we develop various deep learning models needed in early-stage drug development.

Materials Discovery
New materials, from stone all the way to silicon, have led to a paradigm shift in the industry and our life. The concept of digital twin is drawing great attention as a useful approach for efficient materials discovery via high-throughput computations. Here, we further accelerate computational materials discovery by incorporating AI and apply it to designing new OLED molecules.

Representative publications

Scaffold-based molecular design with a graph generative model
Chem Sci 11, 1153 (2020)

ACE-Molecule: An open-source real-space quantum chemistry package
J Chem Phys 152, 124110 (2020)

Predicting drug-target interaction using 3D structure-embedded graph representations from graph neural networks
J. Chem. Inf. Model 59, 3981 (2019)

A Bayesian graph convolutional network for reliable prediction of molecular properties with uncertainty quantification
Chem Sci 10, 8438 (2019)

Molecular generative model based on conditional variational autoencoder for de novo molecular design.
J Cheminform 10, 31 (2018)

Efficient prediction of reaction paths through molecular graph and reaction network analysis
Chem Sci 9, 825 (2018)