The software Progenesis QI (v2.3.6275.47962, Nonlinear Dynamics, Waters, Newcastle, UK) was used to generate a list of potentially differentially abundant compounds for each ionization mode (a compound in Progenesis QI is a combination of retention time and m/z ratio deduced from isotopes and adducts ions) for the data corresponding to 2016 and 2017. All possible adducts and automatic processing were selected, with default automatic sensitivity and no chromatographic peak width indicated.

For the positive ionization mode, the number of compounds for the 3 years of harvest was 14776, 2583 of which had experimental MS2 data. Out of these 2583 compounds, 206 respected the statistical criteria fixed. Progenesis QI gave an identification proposal for 197 compounds, but 15 metabolites were in the end identified.

For the negative ionization mode, 19295 compounds were obtained with Progenesis QI, 4350 of which had experimental MS2 data. Out of these 4350 compounds, 446 respected the statistical criteria fixed. Progenesis gave an identification proposal for 396 compounds, but 14 metabolites were in the end identified. There was no signal-to-noise selection, except for the parameter “default” for the sensitivity of peak picking in Progenesis QI.

Alignment and peak picking were done with default parameters, then all adducts between 0 and 50 min and only compounds with available MS2 data were kept for further statistical analysis. Once the statistics had been performed with R, we also took advantage of Progenesis QI plugins to tentatively identify compounds in the databases Pubchem, MassBank, NIST, ChemSpider and ChEBI, as well as in an in-house database.

R99 (v3.6.0 64-bit) was used to normalize the abundances using internal standard and dry weight and to perform a one-way analysis of variance (ANOVA) with genotype as a factor on the abundances of the compounds in the first two harvests data to establish a list of compounds to search in databases. The criteria of compounds’ choice were p-value ANOVA < 0.05 and maximum fold-change >3.

PeakView (v, SCIEX, Concord. ON, Canada) and database Metlin in addition to the other databases available in Progenesis QI were used to perform manual identification checking.

Annotations and identifications were classified in accordance with the levels of Metabolomics Standards Initiative (MSI)100. Compounds in class 1 were identified by comparison with standards analyzed in the same analytical conditions, based on exact mass, retention time, MS2 fragmentation pattern and UV–visible spectrum, compounds in class 2 were identified based on the same criteria by comparison with data in databases and/or literature. Class 3 was assigned to compounds with the same information as class 2 when they allowed only chemical class determination, typically when the molecule identified is a fragment of a bigger not fully determined molecule.

Calculation of fold-changes and average abundance per genotype were performed again on the three harvests for the already putatively identified molecules when harvest 2018 data were available.

The hierarchical clustering of the heatmaps relative to metabolomics, as well as gene expression and proteomics were obtained with Cluster 3.0 (available at: and Java TreeView (available at:

Raw data were deposited in the repository MetaboLights, under the study number MTBLS1803 (

Note: The content above has been extracted from a research article, so it may not display correctly.

Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.

We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.