Curreio and PFN start joint research on drug discovery using deep learning and struct ural analysis
Curreio, Inc. and Preferred Networks, Inc. announced they will start joint research on drug discovery using deep learning and structural analysis in April.
Both companies will combine structural analysis technology using Curreio’s Cryo-EM and molecular design technology based on deep learning of PFN to speed up and improve the acc
uracy of conventional structure-based drug discovery. We aim to create new compounds for a wide range of drug discovery targets, including targets that have been considered difficult
up to date. This is expected to lead to the development of therapeutic agents for diseases for which effective treatment methods have not been established so far.
Structural analysis technology using Cryo-EM is one of the best technologies in the three-dimensional structural analysis of biomolecules, and was awarded the Nobel Prize in Chemistr
y in 2017. In particular, single particle analysis using a Cryo-EM does not require crystallization and could obtain a structure in an aqueous solution, which is a big advantage against X-ray crystallography. It has brought technological innovations for membrane proteins that were difficult to crystallize.
Comparison of molecular structure analysis with conventional electron microscope (left) and Cryo-EM (right)
（Image source: Martin Högbom/The Royal Swedish Academy of Sciences）
Curreio has its own techniques in sample preparation for Cryo-EM, grid preparation, and 3D structural analysis, and has strengths in structural analysis of protein complexes.
PFN aims to put cutting-edge technologies such as machine learning and deep learning to practical use, and in the medical field. PFN has been working on molecular design using deep learning, miRNA analysis, medical image analysis, and so on. Especially in the field of drug discovery, PFN is conducting research and development focusing on protein structure-based drug discovery.
In conventional drug discovery, it took a huge amount of time and cost to find the new drug candidate compound. By using deep learning for molecular design, the drug discovery process can be greatly streamlined and accelerated.
