De Novo Prediction of RNA-Protein Interactions for Discovery of Tissue-Specific Binding Sites

Overview

In eukaryotes, RNA-binding proteins (RBPs) interact with RNA to control post-transcriptional processes, including alternative splicing, polyadenylation, stabilization, localization, and translational regulation. RBPs and RNAs are differentially expressed between tissues, influencing the transcriptome-wide RNA-protein interactions that occur within each tissue. Current experimental techniques to identify RBP binding sites in vivo, such as UV cross-linking immunoprecipitation and sequencing (CLIP-seq), are expensive, time-intensive, and difficult to perform reproducibly. Here, I introduce a de novo method to predict tissue-specific RBP binding profiles from an RNA sequence in silico. This approach simulates RNA-protein interactions using a probabilistic model that leverages binding affinities predicted from in vitro RNA Bind-n-Seq data and protein compositions measured by mass spectrometry. I apply this method to investigate the binding of 86 RBPs across 32 human tissues, outperforming existing computational methods. I show that the model recapitulates well-studied examples of tissue-specific binding, such as the brain-specific binding of protein NOVA1. In addition, using this approach, I identify NOVA1 as a potential critical protein in the progression of benign adult familial myoclonic epilepsy (BAFME), as NOVA1 binds the UUUCA repeat expansions present in the intronic region of SAMD12. Predictions from this model may aid researchers in understanding the functions of RBPs and uncovering their roles in understudied diseases.

Competition history

• JSHS 2022 Category not listed

Resources

• National JSHS abstract book ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.