30Computational prediction of miRNA signature related to valvular heart disease with atrial fibrillation using coherent data sources at different molecular levels. (15th July 2014)

• Record Type: Journal Article
• Authors: Feng, W; Rao, N; Li, BY; Liu, DY; Yang, F; Liu, HM; Chen, X
• Is Part Of: Cardiovascular research. Volume 103(2014)Supplement 1
• Page Start: S4
• Page End: S4
• Publication Date: 2014-07-15
• Subjects: Cardiovascular system -- Diseases -- Periodicals; Cardiovascular system -- Periodicals; 616.1
• DOI: 10.1093/cvr/cvu077.2
• Languages: English
• ISSNs: 0008-6363

Abstract

Abstract: Purpose: Patients who have both atrial fibrillation and valvular heart disease (AF – VHD) are at a greater (17.5-fold) risk for stroke. MiRNAs, a class of short noncoding RNA controlling gene expression, participate in many physiological processes. This study takes the advantage of published data and results for a short-cut leading to rapid identification of AF-VHD-regulating miRNAs. Methods: A novel method is developed to screen AF – VHD - specific miRNA signature by combining miRNA and gene expression data, which are strong coherent. (i) In both data, the disease and control groups were respectively patients of chronic AF on a background of mitral/aortic VHD and sinus rhythm with VHD; (ii) The average ages of patients with AF – VHD were 72±3 in miRNA data and 74±1 in gene data. For the miRNA expression data with unbalanced sample number, an asymmetric principal component analysis - Bootstrap algorithm is designed to identify differentially expressed (DE) miRNAs associated with AF – VHD. For the identified DE genes related to AF – VHD from the gene expression data, TargetScan and PicTar are used to predict miRNAs that act on these DE genes. The common miRNAs are extracted from two groups of predicted miRNAs, based on receiver operating characteristic curves (ROC) analysis of which, AF – VHD - specific miRNA signature is finally screened. The combined use of two kinds of coherent data is an effective way to avoid the one-sidedness of results and reduce false positives of results, which were probably caused by a single kind of data. Results: A 45-miRNA AF-VHD-specific signature is screened from two coherent data. The ROC analysis shows that the combinations among all the 45 miRNAs have great classification power between AF-VHD and control samples. 21 of 45 miRNAs are verified on miR2Disease to be associated with the inducing diseases of AF –VHD. Compared with the two previous results, 15 of 45 miRNA are same, while 12 of 30 different miRNAs are associated with the inducing diseases of AF –VHD and 12 of the remaining 18 different miRNAs have good discrimination power individually. Therefore, the AF –VHD signature, we screened, has reliability, specificity and diagnostic value of AF – VHD. Conclusions: This study presents a novel method that is suitable for combining the coherent data at different molecular levels to screen a disease – specific miRNA signature. A 45-miRNA AF-VHD-specific signature is warranted. The obtained results demonstrate the potential of miRNAs for the diagnosis of AF-VHD and provide some new insight into potential therapeutic targets of AF – VHD.