CASE

A collaboration with Nippon Shokubai Co., Ltd.

Efficient molecular design of a reactive diluent (acrylic monomer)

AI Drug Discovery / Materials Informatics

Project Background

Nippon Shokubai develops reactive diluents as blending materials for inks. When these reactive diluents are mixed with inks, they are designed to reduce the ink’s viscosity.
With laws and regulations becoming more stringent in each country, it is necessary to ensure that the use of such materials causes less skin irritation.

elix × 日本触媒

Project summary

It would be very time-intensive and expensive to synthesize candidate chemical compounds and evaluate them for use, considering the vast and incomprehensible number of drug-like chemical compounds is estimated to be 1060 compounds.
In this project, we use deep learning to efficiently generate molecules to produce reactive diluents that pose a low risk of skin irritation.

分子設計

Results

  • 01

    Using the predictive model to predict properties, a higher accuracy was achieved than those offered by any currently existing models, particularly in terms of skin irritation.

  • 02

    Using the generative model, we generated a candidate molecule that has an acrylic backbone, which is expected to exhibit low viscosity and lead to low skin irritation.

<GRAPH NEURAL NETWORK>