Introduction: Endometriosis, a complex and common gynecological disorder affecting 5-10% of reproductive-age women, is characterized by the growth of endometrial tissues outside of the uterine cavity. About 176 million women worldwide suffer from endometriosis. The current diagnostic process is laparoscopy, which is a risk-associated and costly invasive procedure. On average it takes 10 years to reach a diagnosis of endometriosis. As such, early intervention is crucial to manage the disease and reduce adverse effects.
Methods: Both DNA-methylation data and RNA-seq data have the potential to uncover molecular mechanisms of diseases. Tissue samples were collected from 80 patients consisting of control and endometriosis patients. The tissues were processed for enrichment-based DNA methylation and RNA-seq data collection. We used the generalized linear model for Differentially Methylated Region (DMR) and Differentially Expressed Gene (DEG) detection. The read count data from DNA-methylation and RNA-seq were concatenated into one big data matrix for predictive analytics. We used the Biosigner and the Decision Tree algorithm for integrative multi-omics analysis. Accuracy, sensitivity, and specificity were calculated to demonstrate the predictive performance and leave-one-out cross validation approach was used for this purpose.
Results: The Biosigner algorithm identified KBDBD2 as a biomarker that was previously reported to be associated with the endometrial cancer. The Decision Tree identified NOTCH3, STXBP5 as biomarker of endometriosis. The integrative model using Decision Tree outperformed both the models that were created using DNA methylation data and RNA-seq data, independently. The accuracy, sensitivity and specificity of the integrative model are 86.21%, 76.92% and 93.75%, consequently.
Impact: Application of multi-omics data integration methods allowed translation of the identification of biomarkers that are associated with endometriosis, but also show susceptibility to other diseases such as endometrial cancers.
Organization – University of Missouri – Columbia
AJoshi T, Akter S, Bromfield J, Pelch K, Wilshire G, Crowder S, Schust D, Barrier B, Davis W, Nagel S