Authors: Eric Li, Valentina L. Kouznetsova, Santosh Kesari, Igor F Tsigelny.
Gastric cancer is one of the most prevalent and deadly cancers in the world. One of the biggest factors for this disease is Helicobacter pylori (H. pylori), infecting roughly half of the world’s population. However, there is a limited understanding of the H. pylori infection at single-cell level. In this study, single-cell RNA-Seq datasets from intestinal metaplasia samples were analyzed.
Using bioinformatics methods, the cells were clustered and cell types were identified with cell type specific marker genes. For each cell type, H. pylori infected cells were compared with control cells using statistical analysis in order to find significant genes and pathways. Then, machine-learning (ML) approaches were used to build models to distinguish H. pylori positive and negative cells, and the severity of infection.
It is found that H. pylori infection is linked to an increase in enterocytes and a decrease in pit mucous cells (PMCs). These changes may promote disease progression from gastritis to gastric cancer. Significantly differentially expressed genes and several pathways such as the MHC class II antigen presentation pathway and the PD-1 pathway were identified. The random forest-based models achieved an accuracy of higher than 97% for detecting positivity and severity.
We identified the specific type of the host cells along with signaling pathways related to H. pylori infection and signaling pathways leading to gastric cancer. We demonstrated that ML methods are useful in detection of the affected by H. pylori PMC cells.
View/Download pdf