Multi-omics integration for target discovery in cancer

Abstract

[[draft to improve]]

Cancer is a complex heterogeneous disease and challenges the identification of novel therapeutic targets. The complex relationships between DNA, transcripts and proteins captured by different single-omics approaches often fail to capture cancer complexity. Moreover, the non-linear relationships between DNA mutation, transcripts expression and protein abundances challenges the linear regression models often used in target discovery. Multi-omics integration can provide a better understanding of the heterogeneous nature of cancer and provide novel therapeutic targets.

Here, I leverage public datasets from The Cancer Genome Atlas program, TCGA. I will focus my efforts on Kidney Cancers: kidney chromophobe (KICH), kidney renal papillary cell carcinoma (KIRP), and kidney renal clear cell carcinoma (KIRC).

First, we gonna explore how molecular differences between these three kidney cancers are captured and reflected in different single-omics layers (i.e.: transcriptomics, proteomics), and how linear multi-omics integration strategies deepen the biological insights.

I will then dig into kidney renal clear cell carcinoma (KIRC) and develop a Deep Learning classifier to discriminate between the different KIRC consensus molecular subtypes. I will compare my method with existing molecular signatures based on transcriptomics data, and test if the new classifier can better classify samples from another cohort and best predict the clinical outcome of the patients.

Finally, I will try to identify novel therapeutic targets based on my molecular subtypes.