Elevated sequence similarity between sibling RRMs in human RBPs. (A) Workflow of the analyses. The human proteins containing RRMs were obtained from the InterPro database, and the RRM sequences were extracted according to the consensus annotations from three different databases. After filtering out the duplicated sequence, 453 RRMs from 391 unique RBPs were analyzed through sequence comparison. (B) Similarity scores between all RRM pairs in human RBPs. Each RRM was aligned with all other 452 RRMs, where the distribution of similarity score is represented by a gray vertical line spanning the mean ± 1× standard derivation. The similarity score between sibling RRMs was represented as a red circle. The order of RRMs along the x-axis is arbitrary. (C) The cumulative frequency of similarity scores between sibling RRM pairs in proteins with 2, 3, 4, or 5 RRMs. As a control, we randomly selected 1000 RRM pairs and computed the cumulative frequency of their similarity scores. (D) Sibling RRMs are more conserved than the shuffled pairs. The histograms of similarity scores between sibling RRM pairs from 112 RBPs that contain two RRMs were plotted (open boxes). As controls, we shuffled the order of these RRMs to generate a simulated set of 112 RBPs with matched sequence composition. The shuffle was repeated 1000 times with replacement, and the mean similarity scores of RRM pairs were plotted as filled boxes. (E) Same as panel D, except 44 RBPs with three RRMs were analyzed.
