Prediction of Potential MicroRNA–Disease Association Using Kernelized Bayesian Matrix Factorization

• Published in: Interdisciplinary sciences : computational life sciences Vol. 13; no. 4; pp. 595 - 602
• Main Authors: Toprak, Ahmet, Eryilmaz Dogan, Esma
• Format: Journal Article
• Language: English
• Published: Singapore Springer Singapore 01.12.2021; Springer Nature B.V
• Subjects: Bayesian analysis; Biomedical and Life Sciences; Breast; Case studies; Colon; Computational Biology/Bioinformatics; Computational Science and Engineering; Computer Appl. in Life Sciences; Computer applications; Disease; Factorization; Health Sciences; Life Sciences; Mathematical and Computational Physics; Medicine; MicroRNAs; miRNA; Neoplasms; Non-coding RNA; Nucleotides; Original Research Article; Pathogenesis; Similarity; Statistics for Life Sciences; Theoretical; Theoretical and Computational Chemistry; miRNA; miRNA–disease relationship; Disease; Similarity measure
• ISSN: 1913-2751; 1867-1462
• DOI: 10.1007/s12539-021-00469-w

MicroRNA (miRNA) molecules, which are effective in the formation and progression of many different diseases, are 18–22 nucleotides in length and make up a type of non-coding RNA. Predicting disease-related microRNAs is crucial for understanding the pathogenesis of disease and for diagnosis, treatment, and prevention of diseases. Many computational techniques have been studied and developed, as the experimental techniques used to find novel miRNA–disease associations in biology are costly. In this paper, a Kernelized Bayesian Matrix Factorization (KBMF) technique was suggested to predict new relations among miRNAs and diseases with several information such as miRNA functional similarity, disease semantic similarity, and known relations among miRNAs and diseases. AUC value of 0.9450 was obtained by implementing fivefold cross-validation for KBMF technique. We also carried out three kinds of case studies (breast, lung, and colon neoplasms) to prove the performance of KBMF technique, and the predictive reliability of this method was confirmed by the results. Thus, KBMF technique can be used as a reliable computational model to infer possible miRNA–disease associations.