Integrating single-cell analysis and machine learning to create glycosylation-based gene signature for prognostic prediction of uveal melanoma

Abstract

BackgroundIncreasing evidence suggests a correlation between glycosylation and the onset of cancer. However, the clinical relevance of glycosylation-related genes (GRGs) in uveal melanoma (UM) is yet to be fully understood. This study aimed to shed light on the impact of GRGs on UM prognosis.MethodsTo identify the most influential genes in UM, we employed the AUCell and WGCNA algorithms. The GRGs signature was established by integrating bulk RNA-seq and scRNA-seq data. UM patients were separated into two groups based on their risk scores, the GCNS_low and GCNS_high groups, and the differences in clinicopathological correlation, functional enrichment, immune response, mutational burden, and immunotherapy between the two groups were examined. The role of the critical gene AUP1 in UM was validated through in vitro and in vivo experiments.ResultsThe GRGs signature was comprised of AUP1, HNMT, PARP8, ARC, ALG5, AKAP13, and ISG20. The GCNS was a significant prognostic factor for UM, and high GCNS correlated with poorer outcomes. Patients with high GCNS displayed heightened immune-related characteristics, such as immune cell infiltration and immune scores. In vitro experiments showed that the knockdown of AUP1 led to a drastic reduction in the viability, proliferation, and invasion capability of UM cells.ConclusionOur gene signature provides an independent predictor of UM patient survival and represents a starting point for further investigation of GRGs in UM. It offers a novel perspective on the clinical diagnosis and treatment of UM.

Integrating multiple machine learning methods to construct glutamine metabolism-related signatures in lung adenocarcinoma

“…To validate our findings, we downloaded expression profiles from eight GEO datasets, including GSE13213 (n=117), GSE26939 (n=115), GSE29016(n=39), GSE30219 (n=85), GSE31210 (n=226), GSE37745 (n=106), GSE42127 (n=133), and GSE68465 (n=442). To ensure comparability across datasets, all expression data was normalized to transcripts per million (TPM), and batch effects were removed using the “sva” package ( 17 ). Prior to analysis, all data was log2 transformed.…”

Mast cell marker gene signature: prognosis and immunotherapy response prediction in lung adenocarcinoma through integrated scRNA-seq and bulk RNA-seq

“…The clinical characteristics were integrated with the risk score utilizing the R package ‘rms’ to construct a more precise nomogram ( 19 , 20 ), which enhanced the predictability of prognostication. The accuracy of the nomogram was assessed by calibration and ROC curves ( 21 ).…”

Single-cell and bulk RNA sequencing reveal cancer-associated fibroblast heterogeneity and a prognostic signature in prostate cancer

“…We used the Seurat v4.1.3 package to process scRNA-seq data. [20,21] Several quality measures were applied: cells with fewer than 200 features (low quality) or >8000 features (doublets/multiplets) were removed; cells with more than 20% mitochondrial genes or 5% hemoglobin genes were removed (low quality) (Supplementary Fig. 1A-B, http://links.lww.com/ MD/J443).…”