Discovery of salivary gland tumors’ biomarkers via co-regularized sparse-group lasso

In this study, we discovered a panel of discriminative microRNAs in salivary gland tumors by application of statistical machine learning methods. We modelled multi-component interactions of salivary microRNAs to detect group-based associations among the features, enabling the distinction of malignant from benign tumors with a high predictive performance utilizing only seven microRNAs. Several of the identified microRNAs are separately known to be involved in cell cycle regulation. Integrated biological interpretation of identified microRNAs can provide potential new insights into the biology of salivary gland tumors and supports the development of non-invasive diagnostic tests to discriminate salivary gland tumor subtypes.