Elix applied machine learning to predict potential drug candidates against COVID-19

Jul 13 2020 newsrelease

Tokyo, Japan [July 13, 2020]

Elix, Inc. (CEO: Shinya Yuki / headquarters in Chiyoda-ku, Tokyo, hereafter as “Elix”) is a research-oriented technology company specializing in deep learning and machine learning. We used machine learning to screen approved drugs and chemical compounds in clinical trials to identify candidate drugs that are likely to be effective against COVID-19 disease. In this study, the RNA-dependent RNA polymerase (RdRp), which is an important protein involved in the replication of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is targeted and inhibitory activity possessed by the chemical compounds against RdRp is predicted.

As a research-oriented technology company specialized in deep learning and machine learning, Elix’s main partners are predominantly the R&D division of major companies. Elix offers full assistance, from consulting, model-development and model-licensing to their partners. Currently, their focus is AI Drug Discovery / Materials Informatics and Computer Vision. In particular, Elix selected AI drug discovery as the main area of focus after attending Blockbuster Tokyo, an acceleration program for pharmaceutical and medical startups, in 2019.

Research background
COVID-19, a disease caused by the SARS-CoV-2, causes serious harm worldwide and there is an urgent requirement to develop a treatment to stem this global health crisis. However, the development of effective new chemical compounds is a long process and typically requires approximately 10 years or more until approval. Therefore, we believe drug repositioning is a more optimal approach than developing an entirely new drug in the current situation. The purpose of drug repositioning is to investigate new potential for already approved drugs and apply these drugs as treatments for new diseases. As approved drugs have already been tested for clinical safety, drug repositioning allows the bypassing of some stages of drug development. In addition, established manufacturing methods can significantly shorten the time required to develop the drugs. The AI drug discovery team of Elix used machine learning to predict whether candidate compounds could be effective for treating COVID-19.

Experimental method
RdRp enzymes are crucial for making copies of RNA viruses and there is a degree of structural similarity between RdRps of different viruses. Moreover, some chemical compounds were reported to effectively target RdRps of several viruses. Therefore, we decided to apply machine learning to identify drug candidates with potential to act on SARS-CoV-2 RdRp. Machine learning is an approach that learns how to solve problems using existing data. Thus, it is necessary to first prepare the input data pertaining to RdRp inhibitory activity and learn the chemical properties based on this data. We gathered data regarding compounds with reported inhibitory activities against the RdRp of the hepatitis C virus (HCV), poliovirus, dengue virus, and influenza virus.

The machine learning models were trained using the data and then used for screening. The primary targets of the screening were FDA*1 approved drugs and compounds in clinical trials, notably antiviral and anti-inflammatory compounds. Anti-inflammatory compounds were included to find candidates that could inhibit SARS-CoV-2 replication while suppressing virus-induced inflammation. Furthermore, as the three-dimensional structural information of the SARS-CoV-2 RdRp is currently available, we concurrently performed docking simulations.

*1FDA…The U.S. Food and Drug Administration, Government of the United States of America, which regulates the approval, regulation of food, drugs, cosmetics and medical devices, evaluates the effectiveness and validity of its products, regulates clinical trials and policy violations.

Research results
The AI drug discovery team of Elix identified multiple drug candidates that could potentially inhibit the RdRp in SARS-CoV-2.
The candidate drugs include remedesivir and baloxavir marboxil, drugs already known to be SARS-CoV-2 RdRp inhibitors. The FDA has issued an emergency use authorization for remdesivir to treat hospitalized patients with severe COVID-19. Baloxavir marboxil is also currently undergoing clinical trials for the evaluation of its clinical efficacy in treating COVID-19.

Additionally, a number of antiviral drugs approved to treat the HCV were predicted to possess effective antiviral activity against SARS-CoV-2. Multiple models predicted that beclabuvir, an HCV RdRp inhibitor, would exhibit inhibitory activity against SARS-CoV-2 RdRp. The drug is considered to be one of the most promising candidates in the study.

Moreover, the overreaction of the immune system, caused by SARS-CoV-2, is believed to trigger inflammation such as pneumonia. Among anti-inflammatory compounds, betulinic acid and ursolic acid are considered to be effective in treating COVID-19. The drugs mentioned in this study are candidate drugs in the fight against COVID-19. However, experiments and clinical trials are required to verify their clinical safety and efficacy. The drugs should not be used to treat COVID-19 without such verification.

Presently, the findings of this study are posted as preprint to arXiv under the following title: “Predicting inhibitors for SARS-CoV-2 RNA-dependent RNA polymerase using machine learning and virtual screening.”

Comments from Elix CEO Shinya Yuki regarding the project
We started this project to apply our experience in the field of AI drug discovery to aid in the fight against COVID-19. We wish to continue to contribute our knowledge to end the pandemic as soon as possible.

About Elix, Inc.
As a research-oriented technology company specialized in deep learning, the company is focused on AI Drug Discovery / Materials Informatics and Computer Vision. Currently the company offers full assistance, from consulting, model-development and model-licensing to their clients.
Company name: Elix, Inc.
Head Office: Daini Togo Park Building 3F, 8-34 Yonbancho, Chiyoda-ku, Tokyo 102-0081 Japan
CEO: Shinya Yuki
Established: November 4, 2016