RepGene: Auditing Condition-Specific Fusion Biases for Multi-View Gene Representations
- 1BGI Research, Beijing
Abstract
This study evaluates how multi-view gene representations should be compared under a matched contract. Each of 18,460 genes is described by five aligned views: DNA sequence-derived features (D), RNA sequence-derived features (R), protein language-model embeddings (P), curated text knowledge (K), and single-cell expression (X). The contract fixes the gene universe, the 256-D output, label-free outer-training-only fitting, one random-forest probe, and task-equal Macro-F1 over nine registered binary tasks, and varies only the fusion operator across four protocol-defined input states: Core (DNA+RNA+protein), Core+K, Core+X, and Core+K+X. Under these states, the top-scoring operator changes with the available view combination and all task-bootstrap intervals overlap, establishing a condition-dependent rather than universal ranking pattern. An independently configured external RF track provides a separate check of the numerical range, while a gene-to-cell boundary diagnostic shows that gene-level differences attenuate toward a dimension-matched random control under expression-weighted aggregation, as predicted by a dilution identity. Historical log-derived multiplicative-interaction screening is retained only as magnitude-only supplementary evidence because fold-level records are unavailable. The contribution is therefore the matched contract and its audit trail: an evidence-graded protocol for turning fusion claims into comparable, condition-bounded measurements.
Related
- DSDAN: A Dual-Stream Multimodal Log-Flow Fusion Framework for Network Configuration Auditing and Compliance-Risk Detection
- A Physics-Informed Neural Network Surrogate for Patient-Specific Hepatic Arterial Hemodynamics in Yttrium-90 Radioembolization: Network Architecture, Boundary-Condition Enforcement, and Data Efficiency
- CoHyFuse: Condition-wise Hypergraph Fusion with Global Connectome in Task-fMRI
- Disentangling non-specific and condition-specific transcriptomic responses: a framework from heat stress in yeast
- Hierarchical distribution matching enables comprehensive characterization of common and condition-specific cell niches in spatial omics data
BibTeX
@misc{hou2026repgene,
title = {{RepGene: Auditing Condition-Specific Fusion Biases for Multi-View Gene Representations}},
author = {H. Hou and T. Xia and L. Hu and Y. Zhang and Y. Li and S. Fang and L. Cao},
howpublished = {bioRxiv},
year = {2026},
month = jun,
doi = {10.64898/2026.06.11.731512},
url = {https://www.biorxiv.org/content/10.64898/2026.06.11.731512v2},
note = {CC BY},
}
RIS
TY - UNPB
TI - RepGene: Auditing Condition-Specific Fusion Biases for Multi-View Gene Representations
AU - H. Hou
AU - T. Xia
AU - L. Hu
AU - Y. Zhang
AU - Y. Li
AU - S. Fang
AU - L. Cao
PY - 2026
DA - 2026/06/11
PB - bioRxiv
DO - 10.64898/2026.06.11.731512
UR - https://www.biorxiv.org/content/10.64898/2026.06.11.731512v2
AB - This study evaluates how multi-view gene representations should be compared under a matched contract. Each of 18,460 genes is described by five aligned views: DNA sequence-derived features (D), RNA sequence-derived features (R), protein language-model embeddings (P), curated text knowledge (K), and single-cell expression (X). The contract fixes the gene universe, the 256-D output, label-free outer-training-only fitting, one random-forest probe, and task-equal Macro-F1 over nine registered binary tasks, and varies only the fusion operator across four protocol-defined input states: Core (DNA+RNA+protein), Core+K, Core+X, and Core+K+X. Under these states, the top-scoring operator changes with the available view combination and all task-bootstrap intervals overlap, establishing a condition-dependent rather than universal ranking pattern. An independently configured external RF track provides a separate check of the numerical range, while a gene-to-cell boundary diagnostic shows that gene-level differences attenuate toward a dimension-matched random control under expression-weighted aggregation, as predicted by a dilution identity. Historical log-derived multiplicative-interaction screening is retained only as magnitude-only supplementary evidence because fold-level records are unavailable. The contribution is therefore the matched contract and its audit trail: an evidence-graded protocol for turning fusion claims into comparable, condition-bounded measurements.
ER -