COVID-19 has caused more than 2.5 million deaths worldwide and the search for new medications and treatments remains impotence. In relation to this, MIT CSAIL researchers are trying to help and for this they have created a deep learning system that helps predict new drug combinations to fight COVID-19.
One of the major challenges with the coronavirus is the fast moving situation and viral mutations. The data needs time to catch up and yet the virus takes no time to slow down. This led scientists from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) to ask: how can we identify the right synergistic drug combinations for the rapidly spreading SARS-Cov2?
The answer is through machine learning capable of drug assessment. This is based on drug synergy, which often occurs through inhibition of biological targets, (like proteins or nucleic acids).
The deep learning is in the form of a neural network. The platform has identified synergistic drug blends for treating viruses like SARS-Cov2.
The platform developed is capable of learning drug-target interaction and drug-drug synergy to mine new combinations. This involves modelling the interaction between a drug and a set of identified biological targets. Through repeated runs the machine learning learns to understand a drug’s antiviral activity. This is expressed as determining the virus yield in infected tissue cultures.
This overcomes the problem that large proportion of compounds have incomplete drug-target interaction information. The machine learning tool can address this requirement in a far faster way, out performing a conventional data science approach.
As examples of how this works, two new drug combinations were found using the neural network: remdesivir (currently approved by the FDA to treat COVID-19), and reserpine, as well as remdesivir and IQ-1S, which, in biological assays, proved powerful against the virus.
The process also enables the potency of a drug to be optimised and for side effects to be reduced. It is even possible to create combinations of drugs against a wider range of diseases. With this, the research team are moving beyond the coronavirus and they are applying the neural network to look into HIV and pancreatic cancer. To further refine their biological modeling the researchers aim to capture additional information like protein-protein interaction and gene regulatory networks.
The machine learning tool is described in the journal PNAS, with the research titled “Deep learning identifies synergistic drug combinations for treating COVID-19.”
