–
April 23, 2024
Imraan Alas, Pharmaceutical Sciences graduate student (Bugni Research Group), will be defending his PhD research thesis:
Computational Approaches to Natural Product Drug Discovery
In the last several decades, antibacterial resistance has continued to rise. Concurrently, the rate of novel antibacterial compounds discovered and approved has dropped. To combat these trends, the ability to leverage computational methods for the improvement of natural product (NP) drug discovery is of paramount importance (Chapter 1). Historically, our lab has prioritized bacterial strains from under-explored environments compared to traditional NP drug discovery research centers, due to the increased rates of re-discovery for antibacterial NPs derived from over-exploited environments.
In order to rigorously rationalize this prioritization scheme, we analyzed marine bacterial strains belonging to the Micromonosporaceae family for biosynthetic gene clusters (BGCs) related to secondary metabolism, serving as the genomic landscape for potential antibacterial NPs produced (Chapter 2). Through our untargeted collection of 38 marine Micromonosporaceae, we identified a novel genus and several under-studied species, broadly determined prospective NPs that were unique within our dataset, and compared our BGCs against public databases to assess novelty in relation to published literature.
While our methodologies to compare our BGCs within our dataset and against public databases were robust, the ability to leverage information regarding our BGCs against public databases to infer relationships across our BGCs in the form of an easily visualizable tool for the prioritization of BGCs was of interest. We developed the BGC Prioritization Display, which incorporates existing software analyses to generate visualizations that depict BGCs in relation to each other based on their similarity to known BGCs in public databases (Chapter 3). This has allowed for the identification of novel BGCs potentially produced by a given bacterial strain.
Though the previously described work focuses on computational approaches utilizing genomic data, the importance of characterizing NP compounds actively produced in a laboratory setting is equally if not more important. In parallel, we’ve developed a novel methodology to connect intracellular biological functions to NP structural information based on analytical chemistry techniques. By leveraging liquid chromatography tandem mass spectrometry (LC-MS/MS) to generate molecular fingerprints usable as inputs for machine learning models, we developed a proof-of-concept multiclass classification model to disambiguate NP compounds that modulate tubulin assembly based on simulated MS2 spectra (Chapter 4). Overall, these findings describe the benefits of incorporating computational approaches into NP drug discovery for the efficient prioritization of novel and relevant compounds.