iDCF: Interpretable deconvolution of cell fractions via biologically-informed deep learning using scRNA-seq data
PLOS Computational Biology · 2026
DOI: 10.1371/journal.pcbi.1014727 ↗Kurz gefasst
Fachartikel in der Zeitschrift PLOS Computational Biology, 2026. Verfasst von Hongjiang Guo und 9 weiteren Autor:innen. Beteiligte Einrichtung: School of Computer Science and Technology, Soochow University, Suzhou, Jiangsu, China. Hauptthemen laut MEDLINE: Deep Learning, Computational Biology, Single-Cell Analysis, RNA-Seq. Die Arbeit nennt 73 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Jiangsu Major Science and Technology Special Fund for Innovative Biologics, Basic Research Program of Jiangsu.
Abstract (Originalsprache)
Precise resolution of cellular heterogeneity within complex tissues is fundamental to deciphering disease etiologies from bulk transcriptomic profiles. While computational deconvolution offers a scalable alternative, current deep learning methods predominantly operate as “black boxes,” neglecting the structural constraints of biological laws. This reliance on purely data-driven feature extraction often yields biologically incoherent predictions and limited mechanistic interpretability. iDCF ( I nterpretable D econvolution of C ell F ractions) is a novel framework that enforces biological topology onto deep neural networks. The iDCF architecture employs a dual-stream design, synergizing a standard deep network with a knowledge-based sparse neural network (KSNN) explicitly masked by pathway definitions and protein-protein interaction (PPI) networks. In comprehensive benchmarks, iDCF achieves top-tier performance, consistently ranking among state-of-the-art methods in accuracy and robustness. iDCF integrates the SHapley Additive exPlanations (SHAP) framework, bridging the gap between computational inference and biological intuition. The model’s decision logic is governed by established biological mechanisms rather than spurious statistical correlations, validating its reliability. Validations across clinical contexts, including Alzheimer’s disease, ovarian cancer, and diabetes, demonstrate iDCF’s ability to recover disease-relevant cellular dynamics. iDCF offers a high-performance, interpretable, and biologically grounded tool for deconvolving cell-type proportions, facilitating deeper insights into tissue heterogeneity in health and disease.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- School of Computer Science and Technology, Soochow University, Suzhou, Jiangsu, China.
- Biomedical Basic Research Center (BBRC) of Jiangsu, Suzhou, Jiangsu, China.
- Jiangsu Key Laboratory of Drug Discovery and Translational Research for Brain Diseases, Soochow University, Suzhou, Jiangsu, China.
- MOE Key Laboratory of Geriatric Diseases and Immunology, School of Basic Medical Sciences, Suzhou Medical College of Soochow University, Soochow University, Suzhou, Jiangsu, China.
ORCID-Profile: Tingfang Wu
Förderung
- National Natural Science Foundation of China
- Jiangsu Major Science and Technology Special Fund for Innovative Biologics
- Basic Research Program of Jiangsu
Zitieren
APA
Guo, H., Wu, T., Wang, W., Jiang, Y., Li, G., Nie, L., Jia, Y., Quan, L., Huang, M., & Lyu, Q. (2026). iDCF: Interpretable deconvolution of cell fractions via biologically-informed deep learning using scRNA-seq data. PLOS Computational Biology. https://doi.org/10.1371/journal.pcbi.1014727
Vancouver
Guo H, Wu T, Wang W, Jiang Y, Li G, Nie L, et al. iDCF: Interpretable deconvolution of cell fractions via biologically-informed deep learning using scRNA-seq data. PLOS Computational Biology. 2026. doi:10.1371/journal.pcbi.1014727
BibTeX
@article{Guo2026,
author = {Guo, Hongjiang and Wu, Tingfang and Wang, Wenzheng and Jiang, Yelu and Li, Geng and Nie, Liangpeng and Jia, Yunhua and Quan, Lijun and Huang, Moli and Lyu, Qiang},
title = {iDCF: Interpretable deconvolution of cell fractions via biologically-informed deep learning using scRNA-seq data},
journal = {PLOS Computational Biology},
year = {2026},
doi = {10.1371/journal.pcbi.1014727}
}
EUVIMED