Script and DataSet for Genome-scale metabolic models

OTUs were mapped to metabolic reactions via a previously published collection of genome-scale metabolic models (GSMMs)140. We included OTUs that could be mapped to a taxonomic rank of family or lower and to at least one GSMM. To analyze metabolic reaction differences between patient groups, a two-sample t-test was used to compare the mean abundances of each reaction and each binary patient classification of interest (e.g., UC vs. CD) between the two groups. The Benjamini-Hochberg procedure141 was applied to correct for multiple testing and adj-p-value ≤ 0.05 between groups was considered significant. The metabolic subsystem classification of reactions was obtained from the GSMMs and for each t-test comparing groups of patients and each subsystem, Fisher’s exact test was performed to determine if the subsystem was overrepresented among the significantly different reactions; subsystems with p-value ≤ 0.05 were considered enriched. All analyses were performed in MATLAB (R2018a).