Identifying likely targets for SARS-CoV-2 using multimodal data analysis
Researchers identify new strategies and software to analyze multiple types of data together to find effective targets for SARS-CoV-2.
SARS-CoV-2, more commonly called Coronavirus, has been a topic of thousands of research studies in the last 15 months, so one would expect that by now we know how this virus works and how to treat COVID-19, the disease it causes. But that is not the case. Despite the volume of experiments and data made available, the complex biology of coronavirus SARS-CoV-2 is not yet fully understood.
Because the biology of this virus is very complex, researchers are still working to identify drug target molecules. Dr Sonika Tyagi, Department of Infectious Diseases, is senior author on a paper published 24 May. She said, "We are a fair way off from developing a drug that can cure the COVID-19 disease. What we do know is that a new drug will need to stop the function of the virus, which is usually through a protein directly linked with the disease."
With colleagues from the University of Melbourne, Dr Tyagi developed new strategies and software to analyze multiple types of data together to identify effective targets.
Dr Tyagi said that these data are of multiple types (or “multimodal”) and highly complex, and sophisticated machine learning methods are needed to extract relevant information from otherwise meaningless information or “noise”.
"Existing studies have focused on analysing a single data type that only captures changes in a small subset of the molecular distress in the body caused by the virus. We propose new strategies that can provide a systematic understanding of the disease, and capture a holistic view of the molecular mechanisms of SARS-CoV-2.
"With our University of Melbourne colleagues, we have integrated multimodal data to find the interconnection between data features. Strongly correlated features were used to identify high confidence drug-targets."
Their approach to analysing multimodal data in a parallel fashion to highlight the interrelationships of disease-driving biomolecules is available as an open-access tool for the research community to reuse and apply on future datasets.
Dr Sonika Tyagi can be contacted on firstname.lastname@example.org
Tyrone Chen, Melcy Philip ,Kim-Anh Lê Cao and Sonika Tyagi. A multimodal data harmonisation approach for discovery of COVID-19 drug targets. Briefings in Bioinformatics. https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbab185/6279836