A computer-based platform for drug discovery has been developed by researchers at The University of Texas at Dallas and Novartis Pharmaceuticals Corporation. This platform has the potential to make the drug discovery process more effective, more efficient, and less expensive.
Dr. Baris Coskunuzer, a professor of mathematical sciences at UT Dallas, and his colleagues developed a method that is based on topological data analysis in order to screen thousands of possible drug candidates virtually and narrow the compound candidates down to those that are most suitable for laboratory and clinical testing. This method is described below.
The findings that were discovered by the researchers will be presented at the 36th Conference on Neural Information Processing Systems, which will take place in New Orleans from November 28 to December 9.
In the early stages of drug development, researchers typically focus on locating a biological target, such as a protein linked to a disease of interest. This is one of the most important steps in the process. The subsequent stage is to examine chemical compound libraries that contain thousands of possible compounds that might be useful or could be modified to have an effect on the target in order to alleviate the cause of the disease or its symptoms. The candidates with the most potential go to the next stage, which is a time-consuming and costly process consisting of laboratory testing, clinical testing, and regulatory approval.
The process of finding a new medication can take ten to fifteen years and one billion dollars to complete. Pharmaceutical companies are looking for a method that is less expensive to accomplish this goal. They want to begin the process by identifying the compounds that have the greatest potential right away so that they do not waste time investigating dead ends.
We have developed an entirely new way of virtual screening that is both computationally efficient and ranks chemicals according to the likelihood that they will be effective.
Professor of Mathematical Sciences at the University of Texas at Dallas, Dr. Baris Coskunuzer
Coskunuzer stated that the strategy taken by his group greatly outperforms other state-of-the-art approaches on big data sets, despite the fact that the virtual screening of libraries of chemical compounds is not a novel concept.
The researchers from UTD and Novartis conceptualised the process of virtual screening as a novel kind of topology-based graph ranking issue, which originates from a subfield of mathematics known as topological data analysis. Their approach identifies each molecular compound by the form of its underlying physical substructure, often known as its topology, in addition to a number of attributes that are both physical and chemical that are possessed by the molecule’s constituent parts. The researchers use this information and create a one-of-a-kind “topological fingerprint” for each chemical. This “topological fingerprint” is then used to rank the compounds according to how well they suit the desired properties.
According to Coskunuzer, “the advantage of our algorithm is that it could screen approximately 100,000 compounds in a couple of days, which is much faster than other methods.”
The next stage will be to adapt the method so that it can be applied to the prediction of molecular properties. This would involve assigning a score to a chemical based on how soluble it is in water. In the body of a living person, a drug’s solubility can be an important factor in determining how well it works.
“If you find a good compound, but it does not have the desired molecular properties—for example, if it is not soluble—then it is likely that it is not going to work. This is because it is highly unlikely that the compound will dissolve. According to Coskunuzer, “you want to be able to evaluate these features initially before a medication candidate gets too far into development.”
Dr. Yulia Gel, a professor of mathematical sciences in the School of Natural Sciences and Mathematics, and Dr. Ignacio Segovia-Dominguez, a postdoctoral research associate in computer science in the Erik Jonsson School of Engineering and Computer Science, are two of the other researchers at UT Dallas who are working on the project.
The AI Innovation Lab at Novartis is home to Dr. Andac Demir, a data scientist, as well as Dr. Bulent Kiziltan, the executive director of the lab. Both of these individuals contributed to the article. Dr. Yuzhou Chen, who received his MS from Temple University in 2017, is currently an assistant professor of computer and information sciences there.
Grants from the National Science Foundation, the Simons Foundation, and the Office of Naval Research are what make it possible for the researchers at UTD to continue their work.
