Meta-analysis of Human Genome for Gastric Cancer Detection
Authors: Priyal Tripathi, Anshika Chauhan, Dr. Jyotsna Talreja Wassan, Dr. Veena Ghuriani
Paper presented at the Equinox 4th Annual International Conference on multi-study genomic meta-analysis for early detection of gastric cancer.
Conference Presentation Details
- Conference: Equinox: 4th Annual International Conference
- Dates: September 4–6, 2024
- Location: Maitreyi College, University of Delhi, New Delhi
- Role: Paper Presenter
Abstract & Summary
Gastric cancer remains a leading cause of oncological mortality worldwide, largely due to late-stage diagnosis. This presentation reported a meta-analytic approach to curate and pre-process RNA-seq human genomic datasets from NCBI GEO and TCGA.
By working with multiple datasets and supervised classification models, we identified consensus transcriptional alterations in gastric adenocarcinoma tissue compared to adjacent normal mucosa. We found that while XGBoost gave the best results (AUC value of 99% and accuracy of 94%, with precision to be 95%) on some of the gastric cancer datasets, chi2 feature selection method showed more robust and promising results over the other gastric cancer datasets. These candidate signatures were subsequently evaluated to assess their predictive diagnostic power.