Comparative evaluation of eight software programs for alignment of gas chromatography–mass spectrometry chromatograms in metabolomics experiments

• Published in: Journal of Chromatography A 2014-12, Vol.1374, p.199-206
• Main Authors: Niu, Weihuan, Knight, Elisa, Xia, Qingyou, McGarvey, Brian D.
• Format: Article
• Language: English
• Subjects: Alignment; Alkanes; Analytical biochemistry: general aspects, technics, instrumentation; Analytical chemistry; Analytical, structural and metabolic biochemistry; Biological and medical sciences; Chemistry; Chromatographic methods and physical methods associated with chromatography; Chromatography; Computer programs; Exact sciences and technology; Fundamental and applied biological sciences. Psychology; Gas chromatographic methods; Gas chromatography–mass spectrometry (GC–MS); Metabolomics; Performance evaluation; Principal component analysis; Principal components analysis (PCA); Retention time alignment; Software comparison; Spectrometry; Spectroscopy
• ISSN: 0021-9673
• DOI: 10.1016/j.chroma.2014.11.005

•Eight alignment programs were evaluated using lab-prepared samples in two trials.•SpectConnect, MetAlign and MetaboliteDetector identified ≥90% of the true positives in Trial 1.•MetAlign and MetaboliteDetector identified ≥80% of the true positives in Trial 2. Since retention times of compounds in GC–MS chromatograms always vary slightly from chromatogram to chromatogram, it is necessary to align chromatograms before comparing them in metabolomics experiments. Several software programs have been developed to automate this process. Here we report a comparative evaluation of the performance of eight programs using prepared samples of mixtures of chemicals, and an extract of tomato vines spiked with three concentrations of a mixture of alkanes. The programs included in the comparison were SpectConnect, MetaboliteDetector 2.01a, MetAlign 041012, MZmine 2.0, TagFinder 04, XCMS Online 1.21.01, MeltDB and GAVIN. Samples were analyzed by GC–MS, chromatograms were aligned using the selected programs, and the resulting data matrices were preprocessed and submitted to principal components analysis. In the first trial, SpectConnect, MetAlign and MetaboliteDetector correctly identified ≥90% of the true positives. In the second trial, MetAlign and MetaboliteDetector correctly identified 87% and 81% of the true positives, respectively. In addition, in both trials >90% of the peaks identified by MetAlign and MetaboliteDetector were true positives.