GO enrichment analysis of clusters
Once targets were detected we ran a gene set enrichment analysis (GSEA)
on the targets of the miRNA in each cluster using the parent-child
algorithm in topGO [v2.46.0 (Alexa & Rahnenfuhrer, 2020; Grossmann,
Bauer, Robinson, & Vingron, 2007)] and a functional annotation for
the longest isoforms of each gene in our P. napi annotation
produced with eggNOG (Rodríguez del Río et al., 2022).. We checked for
enrichment for Biological Processes using the functional annotations of
their identified miRNA targets (Wheat et al., in review ). An FDR
of 0.001 was used as a cutoff, as well as a minimum of 2 members
present. Due to the number of targets in each cluster the results were
dominated by large GO terms that are difficult to interpret. To focus
the GSEA analysis, gene sets within cluster were filtered to only
include those containing a minimal number of miRNA target sites per
gene. An increased number of predicted targets per gene reduces
false-positive detection of miRNA targets (Ritchie, Flamant, & Rasko,
2009), and also indicates coordination of targets within a larger set of
targeted genes. Each cluster had varying numbers of miRNAs targeting
genes, so the minimum number of targeting miRNAs needed per locus to be
included per cluster GSEA was determined to be the number that produced
a majority of the top ten GO terms containing more than one significant
gene.