Artificial Intelligence/Machine Learning (AI/ML) -driven dissection of the microbiome’s multi-factorial impact on colorectal cancer (CRC) immune-landscape to guide clinical immunotherapy.

Most colorectal cancer (CRC)-patients do not respond to immuno-therapy due to immuno-suppressive low-antigenic (MSS) CRC. Previous research highlighted differential microbiome associations with different prognosis values of CRC patients including immunotherapy response. However, current uni-variate/technology-driven approaches do not fully capture the overview of multi-factorial microenvironment context. Specially in CRC, multi-variate AI/ML is needed for immune-state estimation shaped by the gut microbiome interactions and confounding clinico-pathological characteristics. We aim to build a ML model with tumour-immune microenvironment and microbiome data to characterize immune activation/CD8+T cell-exhaustion states in CRC predicting microbiome-driven CRC clinico-pathology and patient survival - extremely unique deliverable with high value for reliable patient prognostic classifiers. We will create an original and versatile analysis pipeline that will be applied on other interlocking multi-omic/multi-dimensional datasets, outlining the high impact potential for this proposal using our unique access to matched tumour scRNAseq data with TCR diversity; microarray tumour-expression data (>500 patients) with matched 16s-RNA; and original clinical trial microarray data with over 6000 CRC patient samples with follow-up. Clearly represents a ‘first-in-class’ innovative immuno-oncological biomarker mining, discovery and validation proposal with clear patient prognostic deliverables.

Keywords:computational immunology, microbiome, immunogenetics, metatranscriptomics, single-cell RNA-seq, classifier

Disciplines:Bioinformatics of disease, Computational biomodelling and machine learning, Computational transcriptomics and epigenomics, Single-cell data analysis