It is well known that viruses make extensive use of the host cell's machinery, hijacking it for the purpose of viral replication and interfere with the activity of master regulatory proteins - including RNA binding proteins (RBPs). RBPs recognize and bind RNA molecules to control several steps of cellular RNA metabolism, such as splicing, transcript stability, translation and others, and recognize their targets by means of sequence or structure motifs. Host RBPs are critical factors for viral replication, especially for RNA viruses, and have been shown to influence viral RNA stability, replication and escape of host immune response. While current research efforts have been centered around identifying mechanisms of host cell-entry, the role of host RBPs in the context of SARS-CoV-2 replication remains poorly understood. Few experimental studies have started mapping the SARS-CoV-2 RNA-protein interactome in infected human cells, but they are limited in the resolution and exhaustivity of their output. On the other hand, computational approaches enable screening of large numbers of human RBPs for putative interactions with the viral RNA, and are thus crucial to prioritize candidates for further experimental investigation. Here, we investigate the role of RBPs in the context of SARS-CoV-2 by constructing a first single-nucleotide \textit{in silico} map of human RBP / viral RNA interactions by using deep learning models trained on RNA sequences. Our framework is based on Pysster and DeepRiPe, two deep learning method which use a convolutional neural network to learn sequence-structure preferences of a specific RBP. Models were trained using eCLIP and PAR-CLIP datasets for >150 RBP generated on human cell lines and applied cross-species to predict the propensity of each RBP to bind the SARS-CoV-2 genome. After extensive validation of predicted binding sites, we generate RBP binding profiles across different SARS-CoV-2 variants and 6 other betacoronaviruses. We address the questions of (1) conservation of binding between pathogenic betacoronaviruses, (2) differential binding across viral strains and (3) gain and loss of binding events in novel mutants which can be linked to disease severity and spread in the population. In addition, we explore the specific pathways hijacked by the virus, by integrating host factors linked to these virus-binding RBPs through protein-protein interaction networks or genome wide CRISPR screening. We believe that identifying viral RBP binding sites will give valuable insights into the mechanisms of host-virus interaction, thus giving us a deeper understanding of the life cycle of SARS-CoV-2 but also opening new avenues for the development of new therapeutics.

RNA-binding proteins (RBPs) are critical host factors for viral infection, however, large scale experimental investigation of the binding landscape of human RBPs to viral RNAs is costly and further complicated due to sequence variation between viral strains. To fill this gap, we investigated the role of RBPs in the context of SARS-CoV-2 by constructing the first in silico map of human RBP-viral RNA interactions at nucleotide-resolution using two deep learning methods (pysster and DeepRiPe) trained on data from CLIP-seq experiments on more than 100 human RBPs. We evaluated conservation of RBP binding between six other human pathogenic coronaviruses and identified sites of conserved and differential binding in the UTRs of SARS-CoV-1, SARS-CoV-2 and MERS. We scored the impact of mutations from 11 variants of concern on protein–RNA interaction, identifying a set of gain- and loss-of-binding events, as well as predicted the regulatory impact of putative future mutations. Lastly, we linked RBPs to functional, OMICs and COVID-19 patient data from other studies, and identified MBNL1, FTO and FXR2 RBPs as potential clinical biomarkers. Our results contribute towards a deeper understanding of how viruses hijack host cellular pathways and open new avenues for therapeutic intervention.