Poster Presentation HUPO 2019 - 18th Human Proteome Organization World Congress

MetaboKit: a comprehensive data processing workflow for DDA and DIA-MS data in untargeted metabolomics (#475)

Hyungwon Choi 1 , Guoshou Teo 1 , Pradeep Narayanaswamy 2 , Stephen Tate 3
  1. National University of Singapore, Singapore, NA
  2. SCIEX, Singapore
  3. SCIEX, Toronto, Ontario, Canada

Compound identification and quantification in untargeted metabolomics has long depended on mass fingerprinting of precursor ions and verification by retention time (RT) matching of standards, forgoing systematic fragmentation of precursor ions and MS/MS scans. This practice was influenced by several major factors: (i) RT is a crucial feature for identification of a compound, but it varies by column types; (ii) MS/MS spectra are variable across instrumentation parameters such as collision energy and the spectra contain only a few fragment ions in many small molecules; and (iii) there often exist isomers with similar RT, adding to the ambiguity that cannot be resolved by MS/MS. All these factors contribute to the difficulty in establishing a universally applicable compound identification workflow.


To address some of these challenges, we developed a software package MetaboKit that embodies comprehensive data extraction workflows for DDA and DIA-MS data processing in metabolomics. MetaboKit has special emphasis on active utilization of MS/MS for both identification and quantification tasks. The first part of MetaboKit performs MS/MS-based compound identification and MS1-based quantification for DDA data, which automatically constructs a library of MS/MS spectra validated by external spectral libraries, each element annotated with in-house retention time for a given chromatographic column. Using the library of spectra, the second part of the tool performs semi- and fully-targeted quantification of precursors and fragments from DIA data. Throughout this process, we account for user-specified adducts and in-source fragments in both identification and quantification steps.


To evaluate the performance, we generated a mixture of 90 metabolite standards and analysed the sample with LC-MS/MS in IDA and DIA mode of scans. We show that MetaboKit has outstanding specificity in the identification and great quantitative accuracy, a feature not demonstrated by other bioinformatics workflows that do not systematically incorporate MS/MS scans in the compound identification.