Fig. 2: Elaboration of gender-specific metabolic patterns using three distinct biostatistical methods. | Communications Biology

Fig. 2: Elaboration of gender-specific metabolic patterns using three distinct biostatistical methods.

From: New perspectives on ‘Breathomics’: metabolomic profiling of non-volatile organic compounds in exhaled breath using DI-FT-ICR-MS

Fig. 2

a Random forest classification with 52 predictors and 500 trees exhibiting an out-of-bag (OOB) error of 0.0099. All males and 49 out of 50 females were accurately classified based on their metabolite profiles, respectively. b Scores plot of a PLS-DA analysis depicting a clear separation of metabolite profiles depending on gender with R2Y = 0.989 and Q2Y = 0.871 for five components. c Volcano plot depicting metabolites with a fold-change greater than 1 and a p-value lower than 0.1 (adjusted using FDR correction) with 570 metabolites significantly differing in binary gender. All calculations and analyses were carried out using MetaboAnalyst 5.0. d VENN diagram representing the data analysis using three different biostatistical tests. Metabolite selection was based on the specific criteria: PLS-DA VIP > 1.5, random forest Mean Dec. Acc. >0.00016, adjusted p-value of <0.1.

Back to article page