Untargeted metabolomic analyses on seeds of six Camelina sativa varieties cultivated for five consecutive years - R Script

See https://doi.org/10.15454/ATTENN, Portail Data INRAE, V1 for a full description of the experiment. Statistical analyses on untargeted metabolomic data Raw data were normalised on the internal standard (Apigenin) and weight of seeds used for the extraction. Normalised metabolites accumulation were expressed as Relative Units (RU). Only metabolites having values higher than 0.05 RU in at least half of the sample number were considered. For these 1624 metabolites, the accumulation was modeled with a linear model to decompose the averaged accumulation by a genotype effect plus a year effect and an interaction between the genotype and the year. Several questions were addressed with this model (i) the differences of accumulation between two genotypes averaged on years or given a year (ii) the differences of accumulation between two years averaged on genotypes or given a genotype (iii) the interactions between genotypes and years by testing if the difference of accumulation between two years in a given genotype is the same for a second genotype. All these questions were written as contrasts in the linear model. Pvalues were adjusted across the metabolites for each question to control the false discovery rate (FDR) and a difference of accumulation for a metabolite was declared significant if its adjusted pvalue was lower than 0.05