Recent advances in the genetic field have seen the emergence of high-throughput experiments, referred to as the '-omics' approaches (such as genomics, epigenomics, proteomics, metabolomics, and transcriptomics). The goal of the omics methods is to provide a richer and unbiased view of the cell and hence offer unique insight into its biological complexity. However, the ongoing flood of a large amount of various genomic datasets that are being generated from these high-throughput experiments is in danger of outstripping our capacity to analyze, interpret the results and better understand the related biology. In fact, such data are often high dimensional, subject to several experimental errors, heterogeneous and may contain both inter-correlated and/or time-varying phenotypes and related subjects. Thus, the challenge for statisticians is to extract relevant biological insight from this vast volume of data these approaches produce and to exploit this insight to guide the development of new diagnostics and therapies for human diseases.
Guided by this principle, the general aim of this research scholars–junior 2 program is to establish a set of suitable multivariate tools that will appropriately deal with data analysis challenges we are facing in the omics era. More precisely, it suggests modern statistical genetic methods that appropriately control experimental errors, provide flexible modeling of multi-level dependence, capture heterogeneity, and deal efficiently with large and complex data. Consequently, the proposed methods will be very useful for analyzing several platforms of omics datasets, such as those resulting from whole genome/exome next-generation sequencing. This may help with maximizing the omics datasets utility by the genomics research community.