PNNL

Abstract

LC-HRMS , as applied to untargeted metabolomics or similar strategies, enables the simultaneous detection of thousands of small molecules in biological specimens, generating complex data sets. Alignment is a crucial step in data processing pipelines whereby mass spectral features derived from common ions across samples are assembled into a unified time-aligned matrix amenable to further analysis. Any factors that limit the reproducibility of liquid chromatography separations complicate the alignment of data. Such factors are particularly prominent when aligning data acquired by different laboratories, generated by non-identical instruments, or between batches of large-scale studies. Previously, we developed a software package called metabCombiner for aligning disparately acquired LC-MS metabolomics data sets. Here we report significant upgrades to metabCombiner that enable stepwise alignment of an unlimited number of untargeted LC-MS metabolomics data sets. To accomplish this, a "primary" feature list is used as a template for matching compounds in multiple "target" feature lists. To demonstrate the new workflow, we aligned four lipidomics data sets from core laboratories generated using each lab's in-house LC-MS instrumentation and methods. We also introduce batchCombine, an application of the metabCombiner framework for aligning multi-batch LC-MS metabolomics experiments. metabCombiner is available as an R package as on Github (https://github.com/hhabra/metabCombiner) and Bioconductor. A new online version, implemented as an R Shiny App, is available at https://karnovskylab.dcmb.med.umich.edu/metabCombiner /.