Community-Centers Identify Robust Biomarkers in High-Dimensional, Low-Sample-Size Gene Expression Data
Abstract
High-dimensional, low-sample-size bulk gene expression data poses a fundamental challenge in transcriptomics, which typically includes tens of thousands of genes but relatively few samples, leading to overfitting and unstable selection for key features as biomarkers. We propose a regression by community centers (RCC) framework tailored for such data. RCC converts gene expression data into a feature proximity network and leverages the phase transition of the giant connected component to identify a critical distance threshold that yields a sparse yet maximally informative network representation, indicated by a minimum normalized shortest compression length. Intuitively, at the criticality, the network balances fragmentation and over-connectivity, allowing meaningful gene-gene associations communities to emerge while filtering noises, such that community-centers capture the most informative and non-redundant signals. These representative genes are then fed to a simple ordinary least squares (OLS) model for downstream predictions. Across five bulk gene expression datasets, RCC displays exceptional prediction performance, clearly outperforming benchmark methods, remains robust to missing data and noises, and does not rely on external biological knowledge. These results suggest that network-based representations provide an effective and general framework for high-dimensional, low-sample-size prediction tasks beyond transcriptomics and illustrate how network-science ideas can support robust learning in data-scarce regimes.
Related
- High Sensitivity Fiber-Coupled Vector Magnetometry using NV Centers in a Femtosecond Laser-Written Diamond Waveguide
- All-fiber quantum current transformer based on nitrogen-vacancy centers
- High-cooperativity coupling and spin-resolved extinction of tin-vacancy centers in a diamond-like microcavity
- Theoretical estimation of nonlinear response function and ultrafast optical spectroscopy of nitrogen vacancy centers
- Electron g-value comparison of paramagnetic muonium and hydrogen centers in rutile TiO2
BibTeX
@misc{li2026communitycenters,
title = {{Community-Centers Identify Robust Biomarkers in High-Dimensional, Low-Sample-Size Gene Expression Data}},
author = {Ruiqi Li and Te Bai and Renaud Lambiotte and Orr Levy and Paul Expert and Daqing Li and Shlomo Havlin},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.00986},
eprint = {2610.00986},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.00986},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Community-Centers Identify Robust Biomarkers in High-Dimensional, Low-Sample-Size Gene Expression Data
AU - Ruiqi Li
AU - Te Bai
AU - Renaud Lambiotte
AU - Orr Levy
AU - Paul Expert
AU - Daqing Li
AU - Shlomo Havlin
PY - 2026
DA - 2026/10/11
PB - arXiv
DO - 10.48550/arxiv.2610.00986
UR - https://arxiv.org/abs/2610.00986
AB - High-dimensional, low-sample-size bulk gene expression data poses a fundamental challenge in transcriptomics, which typically includes tens of thousands of genes but relatively few samples, leading to overfitting and unstable selection for key features as biomarkers. We propose a regression by community centers (RCC) framework tailored for such data. RCC converts gene expression data into a feature proximity network and leverages the phase transition of the giant connected component to identify a critical distance threshold that yields a sparse yet maximally informative network representation, indicated by a minimum normalized shortest compression length. Intuitively, at the criticality, the network balances fragmentation and over-connectivity, allowing meaningful gene-gene associations communities to emerge while filtering noises, such that community-centers capture the most informative and non-redundant signals. These representative genes are then fed to a simple ordinary least squares (OLS) model for downstream predictions. Across five bulk gene expression datasets, RCC displays exceptional prediction performance, clearly outperforming benchmark methods, remains robust to missing data and noises, and does not rely on external biological knowledge. These results suggest that network-based representations provide an effective and general framework for high-dimensional, low-sample-size prediction tasks beyond transcriptomics and illustrate how network-science ideas can support robust learning in data-scarce regimes.
ER -