Network-based integration of cross-dataset proteomic profiles using fold-change directionality.
- 1Graduate School of Frontier Engineering, Kitasato University, Sagamihara, Kanagawa, 252-0373, Japan.
- 2Graduate School of Medical Sciences, Faculty of Life Sciences, Kumamoto University, Kumamoto, Kumamoto, 860-8556, Japan.
- 3Database Division for Life Science, BioData Science Initiative, National Institute of Genetics, Research Organization of Information and Systems, Kashiwa, Chiba 277-0871, Japan.
Abstract
Motivation
The rapid expansion of proteomic data has created new opportunities for large-scale integrative analyses. However, substantial variability across platforms, experimental designs, and processing pipelines limits direct quantitative comparisons among studies. Differential proteomic changes between conditions are often considered to be more reproducible than absolute abundances and may therefore provide a robust basis for cross-dataset integration. However, the systematic ability of differential change-based approaches to capture biologically meaningful relationships across heterogeneous datasets remains unclear.Results
We developed a differential-change framework and applied it to public proteomic datasets. Pairwise contrasts were defined as differential proteomic profiles, and the concordance of up- and down-regulated proteins was quantified using odds ratios. Significant profile pairs were visualized as an integrated network. The treatment of anti-cancer drug doxorubicin vs control (MCF-7) comparison emerged as a central hub, with breast cancer proteome profiles clustering around it and associating with tumor stage (P = 0.0386). Enrichment analysis revealed overrepresentation of lipid- and cholesterol-related pathways.Availability and implementation
The source code for proteome network integration is available at https://github.com/manakanishizaki/proteome-network-integration.git.
Related
- Integrated analysis of transcriptomic and proteomic profiles identifies divergent regulatory networks for intramuscular fat in Chinese versus Western Pigs
- Multilayer network approaches to omics data integration in digital twins for cancer research
- Kinetic asymmetry drives directionality in an ATP-binding cassette transporter
- Reference-signature projection framework for exercise transcriptomic directionality based on the IAV-DI reference axis.
- Systematic comparison of affinity plasma proteomic technologies across multiple cohorts reveals generalizability of proteomic prediction
BibTeX
@article{nishizaki2026networkbased,
title = {{Network-based integration of cross-dataset proteomic profiles using fold-change directionality.}},
author = {Manaka Nishizaki and Norie Araki and Shin Kawano},
journal = {Bioinformatics advances},
year = {2026},
month = oct,
doi = {10.1093/bioadv/vbag321},
url = {https://europepmc.org/article/MED/42959934},
note = {CC BY},
}
RIS
TY - JOUR
TI - Network-based integration of cross-dataset proteomic profiles using fold-change directionality.
AU - Manaka Nishizaki
AU - Norie Araki
AU - Shin Kawano
PY - 2026
DA - 2026/10/23
JO - Bioinformatics advances
DO - 10.1093/bioadv/vbag321
UR - https://europepmc.org/article/MED/42959934
AB - Motivation The rapid expansion of proteomic data has created new opportunities for large-scale integrative analyses. However, substantial variability across platforms, experimental designs, and processing pipelines limits direct quantitative comparisons among studies. Differential proteomic changes between conditions are often considered to be more reproducible than absolute abundances and may therefore provide a robust basis for cross-dataset integration. However, the systematic ability of differential change-based approaches to capture biologically meaningful relationships across heterogeneous datasets remains unclear.Results We developed a differential-change framework and applied it to public proteomic datasets. Pairwise contrasts were defined as differential proteomic profiles, and the concordance of up- and down-regulated proteins was quantified using odds ratios. Significant profile pairs were visualized as an integrated network. The treatment of anti-cancer drug doxorubicin vs control (MCF-7) comparison emerged as a central hub, with breast cancer proteome profiles clustering around it and associating with tumor stage (P = 0.0386). Enrichment analysis revealed overrepresentation of lipid- and cholesterol-related pathways.Availability and implementation The source code for proteome network integration is available at https://github.com/manakanishizaki/proteome-network-integration.git.
SN - 2635-0041
ER -