Related Experiment Video
Updated: Sep 14, 2025
10:05
Single-cell Transcriptomic Analyses of Mouse Pancreatic Endocrine Cells
Published on: September 30, 2018
11.2K
Interpretable machine learning-guided single-cell mapping deciphers multi-lineage pancreatic dysregulation in type 2
Xueqin Xie1, Changchun Wu1, Yuhe Yang1
1Department of Clinical Laboratory, Sichuan Clinical Research Center for Cancer, Sichuan Cancer Hospital & Institute, Sichuan Cancer Center, School of Life Science and Technology, University of Electronic Science and Technology of China, Chengdu, 610054, China.
Cardiovascular Diabetology
|July 24, 2025
Summary
This study decodes pancreatic cell changes in diabetes using machine learning, revealing novel cell markers and altered cell communication. Findings offer potential targets for diabetes management and pancreatic cancer risk stratification.
Area of Science:
- Endocrinology
- Computational Biology
- Genomics
Background:
- Pancreatic cellular heterogeneity is crucial for metabolic regulation.
- Pathological changes in pancreatic cells during diabetes are not well understood.
Purpose of the Study:
- To systematically map pancreatic cellular alterations in diabetes using a machine learning-based single-cell framework.
- To identify novel cell-type-specific markers and understand intercellular communication changes.
Main Methods:
- Integrated single-cell RNA sequencing with machine learning (PanSubPred and PSC-Stat).
- Developed PanSubPred for multi-lineage cell annotation and PSC-Stat for stellate cell activation analysis.
- Analyzed intercellular communication pathways and beta cell heterogeneity.
Main Results:
- Identified 64 cell-type-specific markers (38 novel) with high accuracy.
- Quantified progressive stellate cell activation from diabetes to pancreatic cancer.
- Revealed ductal-centric communication hubs and derived a 15-gene signature for diabetic ductal cells.
- Uncovered diabetes-associated beta cell depletion and expansion of specific subtypes.
- Observed shifts in acinar and ductal cells towards inflammatory and secretory phenotypes, respectively.
Conclusions:
- Presents a machine learning framework for mapping pancreatic cell alterations in diabetes.
- Identified novel signatures, stellate cell dynamics, and beta cell trajectories.
- These findings may offer potential therapeutic targets for diabetes and pancreatic cancer risk stratification.
