Deciphering microbial diversity in tumors by deep learning

About this Class

Our bodies host millions of microorganisms, and our relationship with these tiny living companions is complex. While microbial communities support normal processes and defend against harmful pathogens, infections with some viruses and bacteria can drive and modulate tumor progression. Advances in sequencing technologies and in bioinformatic methods have spurred the discovery of microbes in different cancer types. However, a major bottleneck for the study of microorganisms in human diseases is the difficulty to identify and quantify microbes. Short read sequencing technologies, which are the current standard for microbiome studies, do not support identification of divergent and highly mutated sequences and pose a challenge for correctly mapping reads to diverse microbial genes. We develop deep learning-based sequence analysis frameworks that allow identification of diverse microorganisms in cancers, and minimize reliance on homology-based approaches. Applying these methods to publicly available sequencing cohorts, we detect new viruses that have not been implicated in cancer before and identify microbial proteins that correlate with patients’ outcomes. Dr. Auslander earned her B.S. in computer science and biology from Tel Aviv University and continued her studies in Maryland, where she obtained a computer science Ph.D. from the University of Maryland with a combined fellowship at the National Cancer Institute. She received postdoctoral training at the National Center of Biotechnology Information (NCBI) and joined The Wistar Institute in 2021 as an assistant professor. Speaker: Dr. Noam Auslander from the Wistar Institute.