In Silico Prediction and Validation of Novel RNA Binding Proteins and Residues in the Human Proteome

Deciphering a complete landscape of protein-RNA interactions in the human proteome remains an elusive challenge. We computationally elucidate RNA binding proteins (RBPs) using an approach that complements previous efforts. We employ two modern complementary sequence-based methods that provide accurate predictions from the structured and the intrinsically disordered sequences, even in the absence of sequence similarity to the known RBPs. We generate and analyze putative RNA binding residues on the whole proteome scale. Using a conservative setting that ensures low, 5% false positive rate, we identify 1511 putative RBPs that include 281 known RBPs and 166 RBPs that were previously predicted. We empirically demonstrate that these overlaps are statistically significant. We also validate the putative RBPs based on two major hallmarks of their RNA binding residues: high levels of evolutionary conservation and enrichment in charged amino acids. Moreover, we show that the novel RBPs are significantly under-annotated functionally which coincides with the fact that they were not yet found to interact with RNAs. We provide two examples of our novel putative RBPs for which there is recent evidence of their interactions with RNAs. The dataset of novel putative RBPs and RNA binding residues for the future hypothesis generation is provided in the Supporting Information.