Computational Prediction Of Human Disease-Related MicroRNAs By Path-Based Random Walk

MicroRNAs(miRNAs)have been discovered as important genetic regulation of genes expression in animals and plants.MiRNAs are a class of very little non-coding regulatory RNA molecules that modulate the expression of several genes at the post-transcriptional level and play a critical role in disease pathogenesis.Discovering the relationship between diseases and miRNAs is fundamental for understanding the pathological process of diseases.Therefore,biological examination is a major method of recognizing whether miRNAs are related to any disease.However,this method presented bottlenecks(e.g.time-consumption,and high cost)due to big data from different databases that render it complex.Beforehand,many researchers performed different computational methods to discover relationship between diseases and miRNAs to assist in biological tests.Computational techniques to predict potential disease-related miRNAs is the only immediate way to overcome the presented difficulties.Nevertheless,one main problem for computational methods is the lack of enough bioinformatics methods that predict potential miRNA-disease associations with a high degree of accuracy.Therefore,in this thesis,we propose a computational prediction method of human disease-related miRNAs by path-based random walk to predict potential candidates of disease-related miRNAs to overcome challenges stated in this research area.Moreover,construction of disease-miRNA networks to come up with the similarity between diseases and miRNAs evolved as features extraction between miRNAs and diseases.Based on random walk,the walker moves from each disease's vertex of the disease's network to miRN A vertices by calculating the similarity score between disease-miRNA vertices.As a result,for a given disease with miRNAs,the similarity score has been calculated and the miRNAs with higher similarity scores were confirmed as the potential candidates for a given disease.Lastly,we apply our method to two different types of cancer datasets;Breast Neoplasms,and Pancreatic Neoplasms as our case study to assess the performance of our method.Path-based random walk gave a better prediction accuracy compared to previous models which will contribute to biological investigations and help future researchers to overcome the major issues in this research area.