New York Times is hosting a writing competition where participants get to introduce a cutting edge science or engineering advancement to the public. This is the first time I am submitting to New York Time. My article is attached in the following.
How can computer science accelerate efforts to combat Covid-19?
At first sight, there seems to be no relationship between computer science and the battle with COVID-19. Generally, pharmacology research has focused on developing vaccines that may, in time, control, limit, and eventually eradicate the disease. Computer modeling offers an additional strategy, which has already proven itself in, for example, the fields of climatology and meteorology. So what is on offer?
COVID-19 vaccines stimulate patients’ own immune system to produce antibodies so that most will suffer only mildly or even become immune. However, treatment is also needed for patients who have already been infected. However, gaining approval for new drugs can take years. But there are many drugs originally developed for treating other diseases, which may also prove effective in treating COVID-19. Remdesivir, for example, was originally developed to treat Hepatitis C; now, it is being repurposed to tackle other diseases like Ebola and Marburg Virus. Such drugs have already undergone clinical trials, are approved for use on human beings, and can be quickly deployed.
Drug Target Interaction is one of the best computer science approaches to this kind of drug repurposing research. If a drug and its target virus interact, there is a high probability the drug can be used to cure the disease. Computer models - for example Deep PLA [1] - can calculate this probability by using existing binding data, with an accuracy rate of 82%, and, in test simulations, conduct such analysis for more than 3,000 drugs in an hour.
This approach provides a number of collateral benefits. The most obvious advantage is the elimination of the cost of laboratory experimentation in a bio-secure environment. Another benefit is the encouragement of fruitful collaboration between global experts in related scientific disciplines. However, there are also disadvantages. Although computer modeling can identify potentially useful drugs, clinical trials are still necessary to confirm drug effectiveness. More significantly, the models’ performance in real-world situations is yet not proven, partly because some of the data used to train the model may recur in the test data.
In conclusion, complementing the development of vaccines that can prevent an individual from catching the disease in the first place, computer models can be used to identify potentially effective drugs to treat the symptoms of Covid-19. So why has such a practical and efficient process been neglected? The simple reason is that it is not in the interests of the major pharmaceutical companies to encourage this kind of research. Drug repurposing challenges Big Pharma’s business model - which is based on charging high prices for ‘new’ premium products - by finding inexpensive yet effective ways to use existing products for new objectives. Worse, it risks publication of previously proprietary data, or, worst of all, revealing that a competitor’s product is more desirable. However, the Covid-19 pandemic has reframed the discussion. Collaboration is the order of the day. And collaboration is what computer modeling and computer scientists do best. They deserve generous funding.
[1] https://www.biorxiv.org/content/biorxiv/early/2021/12/03/2021.12.01.470868.full.pdf
