Skip to main content


Fig. 8 | Genome Biology

Fig. 8

From: Insights into the design and interpretation of iCLIP experiments

Fig. 8

A schematic explaining how different extents of cDNA-end constraints affect binding site assignment. a If the iCLIP library contains a broad range of cDNA lengths and unconstrained positions of cDNA-ends, then crosslink sites are identified in an unbiased manner, allowing assignment of the full binding site (RNA map at the bottom). The crosslink sites assigned by cDNA-starts are marked in red bars and a grey bar marks a crosslink site that is incorrectly assigned by a readthrough cDNA. b If cDNA-ends are constrained, most likely as a result of biased RNase cleavage, then the resulting cDNA-starts do not coincide. Nevertheless, if a broad distribution of cDNA lengths is available and the cDNA-ends are placed far enough from the binding site, then crosslink sites can still be identified across the full binding site, allowing correct assignment, as was seen in the case of eIF4A3-iCLIP2 (Fig. 7d). c If cDNA-ends are constrained to a position very close to the binding site, then those cDNAs that truncate at crosslink sites in the 3′ region of the binding site are too short to be isolated and mapped to the genome. Therefore, crosslink sites are identified only in the 5′ region of the binding site, leading to an overly narrow assignment of binding sites, as was seen in some of the sites identified by eIF4A3-iCLIP1 and eIF4A3-iCLIP2 (Fig. 7c, d). d If cDNA-ends are constrained and an iCLIP library contains a narrow distribution of cDNA sizes, then cDNA-end constraints lead to an overly narrow assignment of binding regions, as was seen in the case of eIF4A3-iCLIP1 (Fig. 7c)

Back to article page