74 protein molecular docking.52 In brief, cleaned protein structures were used as receptors, and the peptide core sequence was used to generate 100 different conformers and a global sampling of binding orientations into the peptide binding domain of HLA-II receptors. Following docking, the peptide-HLA-II complexes with the highest complementarity were selected for receptor– peptide refinement in the HADDOCK Refinement Interface.53 Finally, the peptide-HLA complexes were analyzed for the formation of molecular interactions and binding energy using PLIP54 and PRODIGY.55,56 Metagenomic analyses Metagenomic sequencing Metagenomic collection and sequencing has previously been detailed in.57 In brief, participants collected and stored in the freezer their fecal samples directly at home. Fecal samples were collected on dry ice and transferred to the laboratory. Aliquots were stored at - 80°C until further processing. The allPrep DNA/RNA Mini Kit (Qiagen; cat. 80204) was used for DNA isolation. DNA was sent to the Broad Institute (Cambridge, Massachusetts, USA) where library preparation and shotgun metagenomic sequencing were performed on Illumina HiSeq. Metagenomic processing Low-quality reads were discarded by the sequencing facility. Reads aligning to the human genome or to Illumina sequencing adapters were removed using default parameters using the KneadData pipeline (version 0.39). In short, this software uses Trimmomatic58 for adapter removal and quality trimming of reads and Bowtie259 for mapping and removal of reads mapped against the human genome (hg19). Taxonomy abundance estimation was then performed using MetaPhlan3 and default parameters.60 Next, microbial relative abundance was transformed using additive logratios on the relative abundance table (adding 1⁄2 of minimal non-zero relative abundance to each cell in the table), with species geometric mean as denominator (center-log ratio). Bacteria not present in at least 10% of samples were discarded. Microbiome-peptide association analysis Co-occurrence between fecal microbiota and blood antibody–bound peptides was assessed using logistic regression analysis, while adjusting for the effects of age, sex and PhIP-Seq plate in 1,051 participants. In total, we analyzed the relation between 284 bacteria and 2,815 antibodies. Each antibody-bound peptide was modeled in generalized linear models as a response variable in a model including age, sex, PhIP-Seq plate and transformed bacterial abundance as predictors. Microbiome meta-analysis To increase the statistical power to detect associations between gut microbiota and blood antibodies, we combined the results of our cohort with the results derived from the 1000IBD cohort (n = 137, blood and fecal samples collected with <1 year difference) by performing a meta-analysis. We filtered out peptides not seen in at least 10 samples in both IBD and LLD Chapter 3
RkJQdWJsaXNoZXIy MjY0ODMw