Carnegie Mellon’s New Machine Learning Algorithms Comb Natural Products for Pharmaceuticals
The COVID-19 pandemic has given renewed urgency to computational biology, and in particular, the hunt for effective pharmaceuticals. Much of the work to fight SARS-CoV-2 focused on combing through existing databases of known natural and pharmaceutical compounds – and much of it also involved novel, AI-driven techniques for quickly evaluating millions of those compounds. Now, researchers from Carnegie Mellon University (CMU) have developed new machine learning algorithms that hunt for natural products that could serve as effective drugs for diseases ranging from cancers to viral infections.
The algorithms, part of a tool called NRPminer, start with a microbe’s metabolite and genomic signals. Then, it associates those signals with known natural products. “Natural products are still one of the most successful paths for drug discovery,” said Bahar Behsaz, a project scientist in CMU’s Metabolomics and Metagenomics Lab and lead author of a paper describing the research, in an interview with CMU’s Aaron Aupperlee. “And we think we’re able to take it further with an algorithm like ours. Our computational model is orders of magnitude faster and more sensitive.”
Specifically, NRPminer looks for non-ribosomal peptides – the namesake NRPs, which are the primary ingredients of a wide suite of antibiotics and oncological drugs, among others. NRPs continually elude researchers, which find the useful ones difficult to sort out from the rest. “What is unique about our approach is that our technology is very sensitive. It can detect molecules with nanograms of abundance,” said Hosein Mohimani, an assistant professor of computational biology and head of the Metabolomics and Metagenomics Lab. “We can discover things that are hidden under the grass.”
To test the algorithms, the team pitted it against a couple hundred strains of microbes known to be vulnerable to a range of natural product-based drugs. NRPminer identified the hundreds of known drugs – as well as four novel natural products that are now considered promising candidates for drug development, and which the research team is already investigating.
“Our hope is that we can push this forward and discover other natural drug candidates and then develop those into a phase that would be attractive to pharmaceutical companies,” Mohimani said. “Bahar Behsaz and I are expanding our discovery methods to different classes of natural products at a scale suitable for commercialization.”
To learn more about the research discussed in this article, read the paper, “Integrating genomics and metabolomics for scalable non-ribosomal peptide discovery,” which was published in the May 2021 issue of Nature Communications.