Differential Expression Analysis for Endometriosis Diagnosis and Pathology

Overview

1 in 10 women are estimated to have endometriosis, an extremely painful disorder characterized by the external growth of endometrial tissue. It is linked to the increased risk of ovarian cancers, yet the only diagnosis options are invasive and inaccessible due to cost/procedure complications. The goal was to identify potential biomarkers, a noninvasive diagnosis method, with differential expression analysis and validate them to predict endometriosis with 90% F-1 score and accuracy. Currently, there are no identified biomarkers for endometriosis because it is variable based on the phase of the menstrual cycle. The GSE6364 microarray dataset, obtained from the NIH GeoDatabase, contained 37 patients labeled by condition and phase of menstrual cycle. Welch’s t-test and k-means clustering were used to filter the dataset. Potential biomarkers were selected from the resulting k-means clustering heatmaps and validated through the support vector machine model (SVM). 13 biomarkers were selected: SCGB1D2, CYP4B1, CRABP2, SFRP4, TMEM119, FAM110C, CD36, SLC26A2, GABPB1-AS1, HNRNPU-AS1, KIAA1211, PPM1L, MRVI1. The biomarkers passed the t-test with p-values >0.01 and showed significant expression differences in the heatmaps. The SVM model resulted in an F-1 score of 90.93% and 88.99% accuracy. Low accuracy is likely due to overfitting, so the procedure should be done with a larger dataset. The biomarkers can be clinically validated. Furthermore, the expression differences were significant in the early-secretory phase, indicating that endometriosis may be related to problems in the phase. This project is the preliminary step in the development of non-invasive diagnostic tools for endometriosis.

Competition history

• AJAS 2025 Category not listed

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• All Biomedical and Health Sciences projects
• AJAS Biomedical and Health Sciences projects
• Every AJAS 2025 project