2.3 Sequence data processing
Primer and adaptor sequences were trimmed using cutadaptimplemented in trimmomatic (27), setting a minimal Phred score of 25. The two fragments of the paired-end reads were merged usingfastq-join with default parameters and processed with the Quantitative Insights into Microbial Ecology (QIIME2 build 8_2020) pipeline. In details, the Divisive Amplicon Denoising Algorithm 2 (DADA2) was used to denoise the reads (28), that were subsequently clustered into operational taxonomic units (OTUs) and taxonomically classified according to the bacterial Greengenes v.13.8 (29) and fungal UNITE version 7.0 reference databases (30). OTUs which were present in over 80% of samples with a minimum abundance of 0.01% were classified as “core OTUs”. Core OTUs were grouped into three categories, based on their presence in HC, in VKC or in both groups.
QIIME2 pipeline was used for generating the rarefaction curves and for calculating the α diversity metrics (diversity of species within an ecosystem) including the Shannon diversity, Simpson index, observed number of OTUs and phylogenetic diversity.
Weighted and unweighted UniFrac distances, Bray-Curtis and Jaccard dissimilarity between samples were calculated and used for investigation of β diversity (diversity of species between ecosystems) through plotting principal coordinate analysis (PCoA). To determine the factors potentially influencing the conjunctival microbiome, bacterial and fungal communities were compared between groups while controlling for the other variables (age, sex, type of VKC, IgE sensitization, type of feeding, previous presence of atopic dermatitis, etc.).