Focus on… Laurent Guyon and gene selection

on the December 17, 2018

Laurent Guyon is a researcher in bioinformatics at The Biology of Cancer and Infection laboratory (UMR of Grenoble Alpes University, CEA and Inserm). In the framework of Data Institute, he is developing methods to identify genes correlated with cancer patient lifetime.
After a PhD in femtosecond laser spectroscopy, Laurent Guyon became an engineer in microscopy and biology data analysis in CEA-Grenoble in 2010. He is now a multidisciplinary researcher in bioinformatics dealing with gene selection and interpretation using the large datasets produced by various omics experiments. He develops methodologies to identify genes mainly dealing with cancer and microRNAs. In particular, he participated in the creation of the website miRViz that allows biologists to interpret their microRNA datasets.

In the framework of Data Institute, Laurent Guyon collaborates with Florent Chatelain and Rémy Jardillier to identify genes correlated with patient survival. Florent Chatelain (Assistant Professor at GIPSA-Lab, Grenoble INP) conducts research on estimation/detection and large scale inference. Rémy Jardillier (BIG, CEA-Grenoble) carries out his PhD project on mathematical methods to identify prognostic biomarkers for cancer.

To identify and select biomarkers, gene expressions are compared to patient survival. Gene expression occurs in each cell of the organism and ultimately ends up with the production of proteins, one of the important macromolecule with an active role in cellular functioning. In the context of cancer, one can monitor the average messenger RNA levels in a biopsy, which is a piece of the tumor. Genes have several expression levels represented by different colors in the graphic. Those molecular information are correlated with the clinical data of anonymous patients. The aim is to confirm that the selected genes can help predict their survival as prognostic biomarkers.
 
 

This interdisciplinary project leads to a first publication that proposes a review of bioinformatics methods to select prognostic biomarker in Biotechnology Journal. After describing the historical methodologies, the three authors present studies that have been performed in the last few years and illustrate their concepts with a renal cancer dataset.

Laurent Guyon, Florent Chatelain and Rémy Jardillier are preparing a new publication about statistical methods. They use real and simulated data to assess the predictive capability of current algorithms.
 
Link for the publication





 
 
 
Published on June 18, 2019