Sign In

Communications of the ACM

ACM TechNews

Deep Learning Helps Predict Drug Combinations to Fight COVID-19

View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
The coronavirus.

Two new drug combinations were found using the new approach: remdesivir and reserpine, as well as remdesivir and IQ-1S.

Credit: MIT Computer Science and Artificial Intelligence Lab

A team led by researchers at the Massachusetts Institute of Technology (MIT) used deep learning to identify drug combinations that can fight COVID-19 by modeling interactions between drugs and known biological targets related to the virus.

The approach identifyied two new drug combinations: remdesivir and reserpine, and remdesivir and IQ-1S.

Said MIT's Wengong Jin, "By modeling interactions between drugs and biological targets, we can significantly decrease the dependence on combination synergy data. In contrast to previous approaches using drug-target interaction as fixed descriptors, our method learns to predict drug-target interaction from molecular structures. This is advantageous since a large proportion of compounds have incomplete drug-target interaction information."

The model is not limited to a single strain of SARS-CoV-2, and the researchers applied their approach to HIV and pancreatic cancer as well.

From MIT News
View Full Article


Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA


No entries found