Bootstrapped Integrative Hypothesis Test, COPD-Lung Cancer Differentiation, and Joint miRNAs Biomarkers

Integrative Hypothesis Test (IHT) has been recently proposed for an integrated study of hypothesis test, classification analysis and feature selection. This paper not only applies IHT to identifying miRNAs biomarkers for the differentiation of lung cancer and Chronic Obstructive Pulmonary Disease (COPD), but also proposes a bootstrapping method to enhance the reliability of IHT ranking on samples with a small size and missing values. On the GEO data set GSE24709, the previously reported fourteen differentially expressed miRNAs have been re-confirmed via one by one enumeration of their IHT ranking, with two doubtful miRNAs identified. Moreover, every pair of miRNAs is also exhaustively enumerated to examine the pairwise effect via the p-value, misclassification, and correlation, further identifying those that take core roles in coordinated effects. Furthermore, linked cliques are found featured with joint differentiation performances, which motivates us to identify such clique patterns as joint miRNAs biomarkers.