BAYESMETAB: TREATMENT of MISSING VALUES in METABOLOMIC STUDIES USING a BAYESIAN MODELING APPROACH

• Activity Number: 309 - Bayesian Modeling in Physical Sciences and Engineering
• Type: Contributed
• Date/Time: Tuesday, July 31, 2018 : 8:30 AM to 10:20 AM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Jasmit Shah* and Guy Brock and Jeremy Gaskins
• Companies: Aga Khan University Hospital and Ohio State University College of Medicine and University of Louisville
• Keywords: metabolomics; missing value imputation; bayesian algorithm; truncated normal; correlation; detection limit

Abstract

With the rise of metabolomics, the development of methods to address analytical challenges in the analysis of metabolomics data is of great importance. Missing values (MVs) are pervasive, and often ignored, yet the treatment of MVs can have a substantial impact on downstream statistical analyses. The MVs problem in metabolomics is quite challenging, and can arise because the metabolite is not biologically present in the sample, or is present in the sample but at a concentration below the lower limit of detection (LOD), or is present in the sample but undetected due to technical issues related to sample pre-processing steps. In this study we propose a Bayesian modeling approach called BAYESMETAB to feature a cohesive and robust modeling structure for MVs in high dimensional metabolomics data. Our model accounts for MVs due to the truncation threshold, as well as other sources of missingness unrelated to true metabolite abundance. Statistical inference and data imputation are performed simultaneously using an MCMC algorithm. A hypothesis testing framework for differential abundance of metabolites between treatment group is considered, and BAYESMETAB is shown to perform better.