DR-375 AI-driven classification of metastatic tumors with MSI/MSS phenotypes: an integrative computational and clinical perspective

We aim to provide insights into cancer progression and therapy responsiveness by combining diverse molecular sequence data using an AI-driven multimodal classifier. We focus on cancer types that have microsatellite instable (MSI) phenotypes, namely the colorectal, stomach and uterus cancers. To improve the current understanding of the molecular basis of cancer progression and its specific characteristics in metastatic tumors, we work on characterising the molecular cancer variants based on genotype, transcriptome and various clinical outcomes. Our current work shows that variations in short tandem repeats (STRs) of human genomes are associated with gene expression changes in tumor samples. We plan to further search for STR variations in tumors, annotating and cataloguing those STR variants that are informative of metastasis in the three cancer types. We will then search for molecular signatures that allow to identify  tumors that are more likely to respond to specific types of therapy. We ultimately aim to develop bioinformatics methods to identify STR markers linked to patients survival and therapy response.

Maria Anisimova, Institute of Computational Life Sciences, ZHAW, Switserland