BEIJING – As the clock ticks on COVID-19’s spread, Chinese biotech companies are hoping to unleash the power of cloud computing and artificial intelligence to find a cure for the virus.
Scientists are aware of the limited window in which they find themselves. Peiyu Zhang, chief scientist at computation-driven Shenzhen Xtalpi Technology Co. Ltd., said by the time a new drug or vaccine candidate is ready for clinical trials, the epidemic may have already passed, making it difficult to recruit patients for the necessary research.
“Drug repurposing is a quick answer to the urgent need for medication as it screens through a limited number of drugs that have already been proven safe for human use and are readily accessible in the market,” he told BioWorld.
Gilead Sciences Inc.’s NUC inhibitor, remdesivir, is an example. Already proven safe in a phase I study in 2015, the drug candidate was originally developed for Ebola but has also showed potential in treating coronavirus infections. It is now undergoing a phase III trial for COVID-19 in China.
Acceleration in play
Scientists at Xtalpi are working to accelerate drug repurposing using new technologies to produce accurate models quickly and share the findings and data with the research community. With the support of its cloud computing platform, computational chemistry calculations were completed rapidly, without the need for AI acceleration.
“In a few days, we created homology models of the new coronavirus and its protein sub-units, and quickly screened through 2,900 FDA approved drugs and over 10,000 Chinese medicine molecules for 3CLPro and PLpro inhibitors, both of which are likely targets for treating the new coronavirus,” said Zhang.
“We found 183 compounds that can potentially be used to treat COVID-19, and shared this list of potential molecules, their binding affinity ranking, as well as the 3D structure models of SARS-CoV-2,” he added. SARS-CoV-2 is the virus that causes COVID-19.
Currently, Xtalpi is working with CROs to conduct in vitro tests to validate the potency of its drug repurposing candidates, such as carilzomib and saquinavir. The firm is also conducting high-accuracy binding mode analysis on prominent treatment candidates such as Gilead’s remdesivir.
The power of AI
The company plans to apply AI to its research more.
“AI can be used to expedite high-accuracy computational chemistry calculations to create accurate 3D structure models of the virus, and simulate molecular-level interactions between the virus and human receptors to understand the mechanism of how the SARS-Cov-2 virus infects the human body,” Zhang explained.
He said with these fundamental questions answered and enough data accumulated, AI can help design antibodies and vaccines more effectively.
Currently, Xtalpi is using a free energy perturbation method workflow, known as XFEP, which it developed internally, to finish the alanine screening of the receptor binding domain between SARS-Cov-2 spike protein and human ACE2 to evaluate possible virus mutations that are more contagious and pathogenic.
“Combining AI, bioinformatics, and computational chemistry, we can evaluate possible mutations of the virus, and design drugs for strands that are potentially more contagious and pathogenic, to limit its future threat,” Zhang noted.
The hopeful scientist said that AI can be a powerful and efficient tool in developing a broad-spectrum anti-coronavirus medicine that can treat the current epidemic and protect people from similar outbreaks in the future.
Xtalpi is not alone in leveraging AI to fight the coronavirus.
Huawei Cloud, an AI-focused cloud computing unit of Chinese telecom giant Huawei Technologies Co. Ltd, collaborated with experts from Huazhong University of Science and Technology, Xi'an Jiaotong University, and the Beijing Institute of Genomics Chinese Academy of Sciences to screen drug candidates for COVID-19.
In one week, the team screened 8,506 drug candidates targeting the new coronavirus 3CL hydrolase and the S protein/ACE2 receptors. They ended up identifying beclabuvir, bictegravir, dolutegravir, saquinavir and lopinavir as possible treatments for the disease.
Researchers from the School of Basic Medicine Sciences at Peking University also used AI to screen 2,674 approved drugs and 1,500 traditional Chinese medicines. They identified ambroxol hydrochloride oral solution as a potential treatment. Other candidates remain undisclosed.
Outside of China, British AI firm BenevolentAI identified JAK inhibitor baricitinib as a potential treatment after it screened AAK1 inhibitors and high-affinity AAK1-binding drugs. The firm did so by using its knowledge graph, which is a large repository of structured medical information that includes numerous connections extracted from scientific literature by machine learning.