Differential Abundance Methods for Large Heterogeneous-Featured Metabolomics Datasets

Metabolomics deals with the identification and quantification of small molecules in biological systems. Frequently, an aim of metabolomics studies is to identify those metabolites that have differential abundances between two or more conditions. Due to a variety of factors, there are frequently metabolite features that will be zero for a large fraction of samples in either/or both sample classes. Previous work has been done to develop statistical methods capable of testing for differential abundances in metabolomics data with large fractions of zeros. These methods are not appropriate for data from matched pair designs, which are expected to become the standard as metabolomics is applied to more and more disease studies. Furthermore, the currently available methods either make simplistic statistical assumptions, or use the simplest method for not making assumptions about the data available, which are not necessarily appropriate. In addition, how these zeros result from peak correspondence and peak assignment ambiguities has not been examined, nor methods to address peak correspondence/assignment ambiguities been developed. In this proposal we will develop a novel semi-parametric method to perform differential abundance analysis for metabolomics datasets with a large percentage of zero abundance metabolites values, as well as assess the impact of peak correspondence/assignment errors/ambiguities in generating zeros and their impact on differential abundance analysis, and develop methods to mitigate the impact of peak assignment ambiguities.