Drug discovery is like working a jigsaw puzzle. In order to be therapeutic, the chemical compounds that make up drug molecules must be fashioned to fit with proteins in our bodies. It’s such a complex and time intensive requirement for a meticulous fit that new drug development is relatively easy.
To help speed up the puzzle fitting process, SMU researchers have come up with SmartCADD. With the help of artificial intelligence, quantum mechanics, and Computer Assisted Drug Design (CADD) techniques this open source virtual tool dramatically accelerates the screening of chemical compounds, dramatically cutting the drug discovery timelines. A recent study published in the Journal of Chemical Information and Modelling shows that SmartCADD can identify promising HIV drug candidates.
“Antibiotics, cancer treatments, antivirals, and there is an urgent search to discover new classes of drugs,” said Elfi Kraka, head of the SMU Computational And Theoretical Chemistry Group (CATCO). Despite the increasing adoption of AI across many fields, scientific research is still shying away from using AI due to its opaqueness and the quality of data used to train AI.” Those concerns are addressed by SmartCADD, which generates in one day what takes other tools days, even weeks, to find the promising drug candidates”.
How SmartCADD works
SmartCADD leverages advanced filtering technologies and explainable AI, alongside deep learning models, to scan big data of chemical compounds to find drug leads. The tool has two main components: The Pipeline Interface and the Philtre Interface (which collect data, run filters from the SmartCADD pipeline). The different stages of chemical compound testing are aided by these built in filters.
The SmartCADD platform was then demonstrated in three different case studies with drugs used to treat HIV, and researchers believe several of the proteins in the virus are promising targets. Data from the MoleculeNet library was used by SmartCADD, which created and searched a database of 800 million chemical compounds and concluded that 10 million could potentially be HIV drugs. To do that, it then used philtres to find compounds that best matched already approved HIV drugs.
“Ultimately, this is a user-friendly virtual screening platform for researchers to build a highly integrated and flexible drug discovery pipeline,” says Lyle School of Engineering assistant professor Corey Clark, his Division’s deputy director of Research at SMU Guildhall. “I think we will continue to push the work forward to extend their capacities in chemistry and machine learning.”
“Drug discovery is a very challenging field, it requires a collaboration by a combination of scientists to be truly successful,” said Madushanka.
Together, interdisciplinary collaboration helps to refine and improve upon the same idea, through a fresh set of perspectives.”
“Interdisciplinary research is absolutely necessary to make grand research discoveries that really do impact the real world,” says Laird.
Reference: Southern Methodist University. AI and quantum mechanics team up to accelerate drug discovery.
