Prediction of microRNA and gene target from an integrated network in chronic obstructive pulmonary disease based on canonical correlation analysis.

BACKGROUND: Chronic obstructive pulmonary disease (COPD) is a complex disorder with a high mortality. The pathophysiology of COPD has not been characterized till date.

OBJECTIVE: To identify COPD-related biomarkers by a bioinformatics analysis.

METHODS: Here, we conducted the canonical correlation analysis to extract the potential COPD-related miRNAs and mRNAs based on the miRNA-mRNA dual expression profiling data. After identifying miRNAs and mRNAs related to COPD, we constructed an interaction network by integrating three validated miRNA-target sources. Then we expanded the network by adding miRNA-mRNA pairs, which were identified by Spearman rank correlation test. For miRNAs involved in the network, we further performed the Gene Ontology (GO) functional enrichment analysis of their targets. To validate COPD-related mRNAs involved in the network, we performed receiver operating characteristic (ROC) curve analysis and Support Vector Machine (SVM) classification on only those mRNAs that overlapped with COPD-related mRNAs of Online Mendelian Inheritance in Man (OMIM) database.

RESULTS: The results indicate that some identified miRNAs and their targets in the constructed network might be potential biomarkers of COPD.

CONCLUSIONS: Our study helps us to predict the potential risk biomarkers of COPD, and it can certainly help in further elucidating the genetic etiology of COPD.