Integrated multi-omics, single-cell and spatial transcriptomic analyses identify an HK2/S100A2/GALNT5-associated epithelial/ductal-like program in pancreatic ductal adenocarcinoma

Pancreatic ductal adenocarcinoma (PDAC) is characterized by marked molecular and epithelial heterogeneity, but the cross-cohort and cellular context of candidate genes remains incompletely resolved. We integrated bulk transcriptomic and proteomic data with patient-grouped machine-learning assessment, independent-cohort expression replication, cross-cohort functional modelling, single-cell RNA sequencing, spatial transcriptomics, CNV-assisted analysis and cell-line qRT-PCR. Transcriptomic and proteomic screening identified 4,701 differentially expressed genes, 531 differentially expressed proteins and 238 genes with concordant directions. Four-algorithm integration prioritized a 21-gene candidate space. Repeated patient-grouped cross-validation yielded mean outer-fold AUCs of 0.990 for XGBoost and 0.993 for random forest. These values represent exploratory internal stability within a prespecified candidate space, rather than unbiased external predictive performance, because feature selection preceded the resampling procedure. In GSE62452, HK2, S100A2 and GALNT5 showed single-gene AUCs of 0.845, 0.745 and 0.891, respectively, while an equal-weight, within-cohort Z-score produced an AUC of 0.895 but did not outperform GALNT5. Across five RNA cohorts, HK2 was associated with glycolysis and hypoxia, S100A2 with a PDAC basal-like/squamous-aligned program, and GALNT5 with mucin O-glycosylation and O-glycan biosynthesis. Single-cell and spatial analyses supported heterogeneous epithelial localization and descriptive CNV-anchored concordance. CopyKAT classified 1,587 of 3,244 epithelial candidate cells as aneuploid, while 1,009 cells remained undefined. qRT-PCR showed higher relative HK2, S100A2 and GALNT5 mRNA levels in PDAC cell lines than in HPNE cells. These findings associate the three genes with complementary epithelial/ductal-like programs in PDAC, but do not establish causal regulation, tumor specificity, independent prognostic value or direct therapeutic targetability.

Authors

Institutions

Publication Details

Journal
Discover Oncology
Published
2026-09-21
DOI
https://doi.org/10.1007/s12672-026-05941-x
Primary Topic
Single-cell and spatial transcriptomics
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
article

Integrated multi-omics, single-cell and spatial transcriptomic analyses identify an HK2/S100A2/GALNT5-associated epithelial/ductal-like program in pancreatic ductal adenocarcinoma

Zeyang Fan, Jialong Chen, Wence Zhou, Yongqing Zhao et al.
Discover Oncology
Single-cell and spatial transcriptomics
article

Integrated multi-omics, single-cell and spatial transcriptomic analyses identify an HK2/S100A2/GALNT5-associated epithelial/ductal-like program in pancreatic ductal adenocarcinoma

Zeyang Fan, Jialong Chen, Wence Zhou, Yongqing Zhao, Letian Ren, Xin Li
article en

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is characterized by marked molecular and epithelial heterogeneity, but the cross-cohort and cellular context of candidate genes remains incompletely resolved. We integrated bulk transcriptomic and proteomic data with patient-grouped machine-learning assessment, independent-cohort expression replication, cross-cohort functional modelling, single-cell RNA sequencing, spatial transcriptomics, CNV-assisted analysis and cell-line qRT-PCR. Transcriptomic and proteomic screening identified 4,701 differentially expressed genes, 531 differentially expressed proteins and 238 genes with concordant directions. Four-algorithm integration prioritized a 21-gene candidate space. Repeated patient-grouped cross-validation yielded mean outer-fold AUCs of 0.990 for XGBoost and 0.993 for random forest. These values represent exploratory internal stability within a prespecified candidate space, rather than unbiased external predictive performance, because feature selection preceded the resampling procedure. In GSE62452, HK2, S100A2 and GALNT5 showed single-gene AUCs of 0.845, 0.745 and 0.891, respectively, while an equal-weight, within-cohort Z-score produced an AUC of 0.895 but did not outperform GALNT5. Across five RNA cohorts, HK2 was associated with glycolysis and hypoxia, S100A2 with a PDAC basal-like/squamous-aligned program, and GALNT5 with mucin O-glycosylation and O-glycan biosynthesis. Single-cell and spatial analyses supported heterogeneous epithelial localization and descriptive CNV-anchored concordance. CopyKAT classified 1,587 of 3,244 epithelial candidate cells as aneuploid, while 1,009 cells remained undefined. qRT-PCR showed higher relative HK2, S100A2 and GALNT5 mRNA levels in PDAC cell lines than in HPNE cells. These findings associate the three genes with complementary epithelial/ductal-like programs in PDAC, but do not establish causal regulation, tumor specificity, independent prognostic value or direct therapeutic targetability.

Discover Oncology
Lanzhou University Second Hospital (CN), Lanzhou University (CN)
Openalex Percentile: Top 18%
Single-cell and spatial transcriptomics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.