Integrated Single‐Cell and Machine Learning Analysis Identifies Fibroblast‐Associated Hub Genes and Potential Therapeutics in Ischemic Cardiomyopathy
Authors:
- 5 June 2026
- journal article
- Published by Wiley in Chemical Biology & Drug Design
- Vol. 107 (6) , e70329
Abstract
Ischemic cardiomyopathy (ICM), a leading cause of heart failure, is characterised by complex cellular heterogeneity and a dysregulated microenvironment. A systematic computational dissection of its molecular mechanisms and a coherent pipeline from discovery to potential therapeutics is currently lacking. We integrated single‐cell RNA sequencing (scRNA‐seq) data from ICM patients with four independent bulk transcriptomic cohorts. A cardiac cellular atlas was constructed, and candidate genes were filtered through differential expression analysis. Subsequently, a benchmark of 127 machine learning algorithm‐feature selection combinations was performed to identify robust diagnostic hub genes. Their functions were validated at single‐cell resolution via UCell scoring, pseudotime trajectory analysis, and virtual knockout perturbations using scTenifoldKnk. The immune infiltration landscape was assessed using CIBERSORT and MCP‐counter. Finally, computational drug repositioning and molecular docking were employed to screen for potential compounds targeting the hub genes. Machine learning identified a core 5‐gene signature ( NPPA, HTRA1, LUM, ASPN , and OGN ) demonstrating excellent diagnostic performance across independent datasets (AUC > 0.83). Single‐cell analysis revealed that these genes were most abundantly expressed in fibroblasts and were consistently upregulated in ICM. Pseudotemporal trajectory analysis illustrated their dynamic expression patterns. Virtual knockout and functional enrichment indicated that four of these genes ( ASPN, HTRA1, LUM, OGN ) significantly perturbed pathways related to the regulation of inflammatory response. Immune profiling revealed increased infiltration of fibroblasts and plasma cells in ICM. Molecular docking identified the compound LDN‐193189 as a potential lead molecule with high predicted binding affinity (binding energy < −9 kcal·mol −1 ) for ASPN, LUM , and OGN . Through multi‐omics integration and computational biology, this study systematically delineates a fibroblast‐centric molecular network involving key hub genes and an altered immune microenvironment in ICM and computationally proposes a potential therapeutic candidate. These findings provide a crucial computational foundation and experimental direction for understanding ICM pathology and developing novel therapeutic strategies.Keywords
This publication has 42 references indexed in Scilit:
- Spatial omics: Deciphering heterogeneity in the tumor immune microenvironment and resistance to immunotherapyCurrent Proteomics, 2025
- Single-cell and machine learning approaches uncover intrinsic immune-evasion genes in the prognosis of hepatocellular carcinomaLiver Research, 2024
- MUTYH is a potential prognostic biomarker and correlates with immune infiltrates in hepatocellular carcinomaLiver Research, 2022
- Targeting MuRF1 by small molecules in a HFpEF rat model improves myocardial diastolic function and skeletal muscle contractilityJournal of Cachexia, Sarcopenia and Muscle, 2022
- scTenifoldKnk: An efficient virtual knockout tool for gene function predictions via single-cell gene regulatory network perturbationPatterns, 2022
- Prognostic value of natriuretic peptides in heart failure: systematic review and meta-analysisHeart Failure Reviews, 2021
- The role of Smad2 and Smad3 in regulating homeostatic functions of fibroblasts in vitro and in adult miceBiochimica et Biophysica Acta (BBA) - Molecular Cell Research, 2020
- Profiling Tumor Infiltrating Immune Cells with CIBERSORT2018
- The pathophysiology of heart failure with preserved ejection fractionNature Reviews Cardiology, 2014
- Plasma active matrix metalloproteinase 9 associated to diastolic dysfunction in patients with coronary artery diseaseInternational Journal of Cardiology, 2011