Correlation of miRNA expression to the number of assigned clusters. Here, miRNAs have been assigned to a cluster when they are among the top 200 expressed miRNAs and match the first seed site downstream of the main cross-linking site. Neither the BCBL1 PAR-CLIP data (a) nor in the BC3 PAR-CLIP data (c) show strong correlation. (b) and (d) illustrate how many seven-mer seeds match to clusters when the top 40,100 and 200 miRNAs are considered and when seeds are searched in the whole cluster (all) and only downstream of the main cross-linking site (xlink). Even the strictest assignment (top 40 xlink) leads to a considerable number of approximately 1,000 ambiguous clusters in both datasets and at the same time to about 80% unassigned clusters. The fraction of unassigned clusters drops below 50% when the top 200 miRNA seeds are searched in the whole cluster but with the cost of having thousands of ambiguous assignments.
Erhard et al. Genome Biology 2013 14:R79 doi:10.1186/gb-2013-14-7-r79