Microbial metagenomic sequencing and metabolites bioinformatics analysis

A total of 10 fecal pellet samples from 10 monkeys were analyzed by metagenomics sequencing. We generated a total of 1,326,695,659 sequence reads with an average of 44,223,189 total reads per sample. The quality of sequencing data were examined using FASTQC (V 0.11.7) and MultiQC (V 1.7). 16 S ribosomal RNA from metagenomic data were filtered by SortMeRNA (V 2.1), and OTU was clustered in the Mothur (V 1.41.1) pipeline. Shotgun metagenomic reads were first trimmed and filtered to the host contamination using Trimmomatic (V 0.36) and Bowtie2as part of the KneadData (V0.6.1) pipeline (https://bitbucket.org/biobakery/kneaddata/wiki/Home). Kraken (V 1.0) and Bracken (V 1.0) was used to classify metagenomic sequences. Metagenomic sequences were assembled using Megahit (V 1.1.3) and QUAST (V 5.0.2) was used to check assembly quality. Coding sequences 40,461,161 were predicted by Prokka (V 1.12) from the metagenomic of the Megahit assembly. Among them, 1,817,945 sequences with amino acid lengths greater than 200 were merged by CD-HIT (V 4.7), and finally, the abundance calculation was performed using the Salmon (V 0.12.0) software to obtain 1,709,060 non-redundant coding sequences. The gene functional annotation was determined through ortholog assignment by eggNOG-mapper (V 1.0.3) (Supplementary Fig. 1).