2.6. Statistical analysis

Binary logistic regressions using IBM SPSS Statistics (version 25) were implemented to analyse the association between microorganisms and types of pneumonia. Linear regressions using IBM SPSS Statistics (Version 25) were used to evaluate associations between the presence of the numbers of viruses and the presence of different bacteria. Specificity, sensitivity and agreement of detection of BHV1, BRSV, BVDV, BCV and BPIV3 between metagenomic sequencing and qPCR were determined using 2 × 2 tables. p < .05 is defined as statistical significance.

Samples were clustered according to the prevalence of pathogens using UPGMA (Unweighted Pair Group Method with Arithmetic Mean), and relationships visualized as a heatmap generated using heatmap.plus 1.3 in R (Williams et al., 2019). The distance matrix was created for hierarchical clustering using hclust function in R. The optimal number of clusters was determined by the elbow method and further confirmed with gap statistics using fviz_nbclust and combining clustering methods using NbClust function in R.