Successful treatment of cancer is still a challenge and this is partly due to a wide heterogeneity of cancer composition across patient population.
Accounting for such heterogeneity is very difficult. Clinical evaluation of tumor heterogeneity often requires the expertise of anatomical pathologists and radiologists.
Partner CURIE participated in a data challenge on matrix factorization and deconvolution methods in Aussois (French Alps) by the end of November. There participants with different backgrounds worked in teams to solve certain challenges.
The challenge was dedicated to the quantification of intra-tumor heterogeneity using statistical methods on DNA methylation and transcriptomic data in cancer. In particular, it focused on estimating cell types and proportions in biological samples (in vivo and in silico mixtures) for which transcriptome and methylome profiles were generated.
The goal was to explore various statistical methods for deconvolution analysis and to compare their performance between transcriptome and methylome data.
Dr. Jane Merlevede, researcher in the Barillot-Zinovyev’s group at Institut CURIE (Paris), participated in the challenge. She presented a poster on the deconvolution of various types of omics data using matrix factorization approaches in the context of iPC with an application to Medulloblastoma.
Jane was successfull in the first challenge and solved it by applying the deconvolution methodology developed at CURIE and based on Independent Component Analysis.
