Home / Science / Deep learning gives a boost to drug design

Deep learning gives a boost to drug design



Deep learning gives a boost to drug design

A calculation tool created at Rice University could help pharmaceutical companies expand their ability to investigate drug safety. Credit: Kavraki Lab / Rice University

When you take a drug, you want to know exactly what it does. Pharmaceutical companies go through extensive testing to make sure you do.

With a new deep learning-based technique created at Rice University̵

7;s Brown School of Engineering, they may soon have better management of the performance of drugs being developed in the human body.

Rice lab of computer science Lydia Kavraki introduced Metabolite Translator, a computational tool that predicts metabolites, the products of interactions between small molecules such as drugs and enzymes.

Rice researchers take advantage of deep learning methods and the availability of huge reaction data sets to give developers a big picture of what a drug will do. The method is not bound by the rules that companies use to determine metabolic reactions, paving the way for new discoveries.

“When you’re trying to determine if a compound is a potential drug, you need to check for toxicity,” Kavraki said. “You want to confirm that it does what it should, but you also want to know what else might happen.”

Research by Kavraki, lead author and graduate student Eleni Litsa and alumna Rice Payel Das of IBM’s Thomas J. Watson Research Center, is detailed in the journal Royal Society of Chemistry Chemical sciences.

Researchers trained Metabolite Translator to predict metabolites through any enzyme, but measured its success against existing rule-based methods that focus on enzymes in the liver. These enzymes are responsible for detoxification and elimination of xenobiotics, such as drugs, pesticides and pollutants. However, metabolites can also be formed through other enzymes.

“Our bodies are webs of chemical reactions,” Litsa said. “They have enzymes that act on chemicals and can break or form bonds that change their structures into something that could be toxic or cause other complications. Existing methodologies focus on the liver because most xenobiotic compounds are metabolized there. With the our job, we’re trying to capture human metabolism in general.

“The safety of a drug depends not only on the drug itself, but also on the metabolites that can form when the drug is processed in the body,” Litsa said.

The rise of machine learning architectures that operate on structured data, such as chemical molecules, makes work possible, he said. Transformer was introduced in 2017 as a sequence translation method that has found extensive use in language translation.

Metabolite Translator is based on SMILES (for “simplified molecular input line entry system”), a notation method that uses plain text rather than diagrams to represent chemical molecules.

“What we are doing is exactly the same as translating a language, like English into German,” Litsa said.

Due to the lack of experimental data, the laboratory used transfer learning to develop Metabolite Translator. They first pre-trained a Transformer model on 900,000 known chemical reactions and then fine-tuned it with data on human metabolic transformations.

The researchers compared the results of Metabolite Translator with those of several other predictive techniques by analyzing known SMILES sequences of 65 drugs and 179 metabolizing enzymes. Although Metabolite Translator was trained on a general non-drug specific dataset, it worked like commonly used rule-based methods that were developed specifically for drugs. But it also identified enzymes that are not commonly involved in drug metabolism and have not been found with existing methods.

“We have a system that can predict just as well as rule-based systems, and we haven’t put rules into our system that require manual labor and expert knowledge,” said Kavraki. “Using a method based on machine learning, we are training a system to understand human metabolism without the need to explicitly code this knowledge in the form of rules. This work would not have been possible two years ago.”

Kavraki is computer science professor Noah Harding, professor of bioengineering, mechanical engineering, and electrical and computer engineering, and director of Rice’s Ken Kennedy Institute. Rice University and the Texas Cancer Prevention and Research Institute supported the research.


Engineers use machine learning to accelerate bioscaffold development


More information:
Eleni E. Litsa et al, Prediction of drug metabolites using neural machine translation, Chemical sciences (2020). DOI: 10.1039 / D0SC02639E

Supplied by
Rice University




Quote:
Deep Learning Gives Drug Design a Boost (2020, October 5)
retrieved 6 October 2020
from https://phys.org/news/2020-10-deep-drug-boost.html

This document is subject to copyright. Apart from any correct behavior for private study or research purposes, no
part may be reproduced without written permission. The content is provided for informational purposes only.




Source link