Figure 2
Workflow for proteomics data analysis. (a) Number of proteins identified by unique FASTA identifier. Proteins with LFQ = 0 for all related samples were filtered out before downstream analysis. (b) Analysis 1 (green outline) included proteins with LC–MS/MS data from Experiment 2 (Pilot) and Experiment 3 (Comprehensive) combined and was more conservatively filtered due to the overrepresentation of proteins with no valid values in the Experiment 2 subset. Analysis 2 (blue outline) used only proteins identified from Experiment 3. Missing values (LFQ = 0) were imputed from a downshifted normal distribution and protein expression between humidity groups was analyzed by Welch’s t-Test. Distribution of the log10(p) and group mean difference (log2 scale) are shown below: black vertical lines represent a mean difference of 0.58 (1.5-fold-change), the red and blue horizontal lines represent p = 0.1 and p = 0.05, respectively. This data was arranged by ascending p-value and assessed by principal component analysis. Separation between humidity groups was observed for the top 95 and 515 proteins for Analysis 1 and Analysis 2, respectively, and these points are indicated on the graphs. Analysis 1 concluded due to the low number of significantly differentially expressed proteins identified. Analysis 2 separated the 515 proteins into those positively (red) and negatively (blue) correlated with the first principal component, and these lists were filtered by p < 0.1. These lists were mapped to available gene names by the UniProt Retrieve ID/Mapping tool and supplied to Metascape for gene enrichment analysis.
