Factominer

Preliminary identification of candidate compounds for omission was performed via PLS1 analysis (Supplementary Table S1) of the descriptive analysis data and volatile profiling data; the whole volatile dataset was used as the x-variable and each aroma descriptor was a y-variable in Unscrambler (CAMO Software AS, Oslo, Norway)^{17 (link)}.
Using the data generated by the three panelists in the current study, the mean aroma intensity for each mixture was calculated, and the total number of times each attribute was identified for each mixture was calculated and reported as frequency counts. Correspondence analysis (CA) on the aggregated, check-all-that-apply datasets and Multiple Factor Analysis (MFA) comparing each panelists’ datasets were performed using the “ca” and “FactoMineR” packages for the R statistical program (R Foundation for Statistical Computing, Vienna, Austria), respectively.

In screening assay, Student’s t-test was used to compare the total gas and CH₄ production levels in the control bottles with those levels in bottles containing a given plant additive from the same incubation run. The effects were expressed as relative change to the value of the control for the specific incubation run. The confirmation assay results were analysed using one-way analysis of variance, followed by Newman–Keuls multiple comparison tests. All statistical analyses were performed using GraphPad Prism, version 5.0 (GraphPad Software Inc., La Jolla, CA, USA), and a p-value < 0.05 was considered statistically significant. To identify bacterial lineages and other parameters that differentiated the control and treatment groups, we performed principal component analysis using the fviz_pca_biplot function in the FactoMineR [39 ] package of R-software, version 4.0.3 (The R Foundation for Statistical Computing, Vienna, Austria). The non-parametric Kendall rank-correlation coefficient was calculated to identify correlations among CH₄ production, fermentation characteristics, bacterial communities, and PSMs using the PROC CORR function in SAS software, version 9.4 (SAS Institute, Cary, NC, USA).

To analyze the high dimensional results obtained with the bias calculations, we performed a standard Principal Component Analysis (PCA) using the R package “FactoMineR” (Ver. 3.6.2, R Foundation for Statistical Computing, Vienna, Austria). Hierarchical clustering was performed based on the PCA results [63 ] using the HCPC function in the R package “FactoMineR” as well as the fviz_dend function in the R package “factoextra”.