L'utilisation des outils bioinformatiques pour caractériser le paysage immunologique du cancer de la prostate
|Advisor:||Droit, Arnaud; Fradet, Yves|
|Abstract:||As part of my PhD, I developed applied data analysis approaches to perform a multi-omic analysis of prostate cancer (CaP). My project was split into two distinct parts corresponding to the two articles integrated into the body of my document. A first part of the work consisted in recovering omics data of different types (RNA-Seq, Methylation, CNA, SNA, miRNA, clinical data) associated with CaP and preparing them with an adapted bioinformatics pipeline. Then, my goal was to seek to highlight new immunity checkpoints associated with biochemical recurrence (BCR) in CaP through these data. To fulfill this objective, I used a special approach based on Principal Component Analysis (PCA) and Partial Least Squares Regression (PLS) algorithms. This has brought out a specific family of immunity checkpoints, the LILR family, which can potentially be a target family in immunotherapy. Second, I used the same data to develop a machine learning (ML) analysis protocol. The aim of this work was to show that it was possible to predict whether or not patients would relapse from RNA-Seq data. I have shown that even with small datasets, one can achieve very good prediction scores and that current ML algorithms take into account the technical variability of the diverse data sources in the CaP. It is therefore possible to use current biobanks owned by research structures around the world to create larger datasets.|
|Document Type:||Thèse de doctorat|
|Open Access Date:||24 May 2021|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.