Review of statistical learning methods in integrated Omics studies (An integrated information science)

Abstract

Integrated Omics is becoming a new channel for investigating the complex molecular system in modern biological science and sets a foundation for systematic learning for precision medicine. The statistical learning methods that have emerged in the last decade for Integrated Omics are not only innovative but also multidisciplinary with integrated knowledge in biology, medicine, statistics, machine learnings and artificial intelligence. Here we review the non-trivial classes of learning methods from the statistical aspects and streamline these methods within the statistical learning framework. We review the statistical methods of exploratory and supervised learning from publications available in scopes and google scholars. We also discuss the computational methods which include a comparison between Bayesian and non-Bayesian approach, the sparsity and regularization approach to dealing with studies of fewer observations than the number of features and studies with sparse connections. Lastly, we provide insights in the meta-analysis method used in the integrated omics science. For the completeness of the review, a summarized table of currently available software for Omics are provided in the supplementary materials.