2.12. PICRUSt2

Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2) analysis was performed from sequencing data, as shown before [15] to predict the metabolic pathways. Demultiplexed and trimmed Fastq sequence files for each sample were curated from Ion Reporter and processed using QIIME2. Quality filtered demultiplexed sequences were denoised by removing chimeric sequences and correcting amplicon errors using the DADA2 plugin. Representative sequences and the feature table generated by DADA2 were used as input files for the PICRUSt2 analysis pipeline. Metabolic pathways were assigned based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) Ortholog (KO) database. Read abundance data for all predicted pathways were converted to relative abundance (%). Data were then processed and analyzed by R software (version 4.0.2) to prepare PCA plots and by the Galaxy server (https://huttenhower.sph.harvard.edu/galaxy/) for LEfSe analysis using LDA score 2.5 as a threshold level.