Prediction Of RNA Secondary Structure And Applications In Liquid Biopsy

RNA molecules take on a vast number of complicated structures,and their binding to proteins further contributes to complex three-dimensional shapes.Accurate detection of RNA secondary structure and RNA-protein interactions can help us understand the regulatory mechanisms of RNA in various biological processes;abnormal regulation of RNA molecules may influence the path of disease progression,making exploration of the role of RNA in tumorigenesis of great significance for clinical diagnosis and treatment.Various RNA secondary structure prediction tools are capable of predicting base pairing relationships at poor accuracy,whereas high-throughput experiments based on chemical or enzymatic probes can detect base pairing state of native RNA in vivo.In view of this,we developed RME,an algorithm that extracts reliable base pairing probabilities from experimental data to assist computational prediction.This algorithm constructs specific distribution models for different experimental methods,and estimates base pairing probabilities with Bayesian posterior probability.RME not only predicts structures at high accuracy,but also facilitates comparison among different experiments.RNA-binding proteins(RBPs)determine the fate of almost all RNA molecules through forming complex post-transcriptional regulatory networks.In order to detect such networks,we collected various CLIP-seq datasets,used non-negative matrix factorization method to perform soft clustering analysis on RNA binding sites of RBPs,and finally obtained various RBP clusters with different synergistic functions.This algorithm helps to explain biological functions of RNA regulatory elements as well.With the discovery of extracellular RNA(ex RNA)in body fluids,ex RNA has been explored as biomarkers for cancer diagnosis and prognosis.In order to extract biological information contained in sparse ex RNA fragments,we defined novel domain features and applied machine learning to identify reliable biomarkers for cancer diagnosis and classification.In summary,our research includes both fundamental and clinical research: optimization of RNA secondary structure prediction algorithm,construction of RBP regulatory networks and identification of reliable RNA biomarkers in liquid biopsy.