Benchmarking Active Spot Selection for Cost-Efficient Spatial Transcriptomics
Abstract
Spatial transcriptomics (ST) measures gene expression in tissue context, but dense capture grids can be costly and may repeatedly sample morphologically similar regions. Most active learning strategies were developed for categorical labels and independent samples. We conduct a retrospective pool-based benchmark of active learning versus uniform Random sampling for ST, where expression vectors are high-dimensional and continuous and candidates are spatially correlated. Using two fully profiled public ST cohorts, we mask candidate expression vectors and simulate multi-round selection with uncertainty-based Monte Carlo dropout (MC-dropout) and temporal output discrepancy (TOD), and diversity-based CoreSet and TypiClust-inspired selection. We compare 160 completed configurations at 5%, 10%, 30%, and 50% of the fold-wide training spot pool under patient-level cross-validation, with a separate full-label reference. Within each budget, strategies share the selection schedule, morphology-to-expression predictor, and optimization protocol. We assess mean per-gene within-slide Pearson correlation coefficient (PCC), expression-cluster agreement, and Moran's I fidelity. On HER2-positive breast cancer, pooled mean PCC differences from Random across the four active strategies were -0.0176, -0.0117, +0.0056, and +0.0057 at 5%, 10%, 30%, and 50%, respectively. On cutaneous squamous cell carcinoma (cSCC), three strategies were below Random at 5%, and all four were below Random at 10%. On HER2-positive breast cancer, CoreSet and MC-dropout had lower PCC but higher expression-cluster agreement than Random at the two smallest budgets; this pattern did not reproduce on cSCC. Under the reported fixed training horizons, the evaluated active strategies do not consistently improve on Random at small budgets, and rankings depend on the evaluation measure.
Related
- Comments on "SpatialDG: a novel spatial domain identification method for spatially resolved transcriptomics data based on dual-graph neural network".
- Spatial transcriptomics reveals microenvironmental heterogeneity in osteoarthritic subchondral bone
- 1170 Spatial transcriptomics identifies immune heterogeneity in estrogen receptor positive postpartum breast cancer
- Efficient differential expression analysis of large-scale single-cell transcriptomics data using Dreamlet
- stFormer integrates spatial ligand signaling into a foundation model for spatial transcriptomics
BibTeX
@misc{zhu2026benchmarking,
title = {{Benchmarking Active Spot Selection for Cost-Efficient Spatial Transcriptomics}},
author = {Zheyu Zhu and Junchao Zhu and Fengbei Liu and Tianyuan Yao and Gelei Xu and John Cannon and Haichun Yang and Yuankai Huo and Mert R. Sabuncu and Ruining Deng},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.27208},
eprint = {2609.27208},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.27208},
}
RIS
TY - UNPB
TI - Benchmarking Active Spot Selection for Cost-Efficient Spatial Transcriptomics
AU - Zheyu Zhu
AU - Junchao Zhu
AU - Fengbei Liu
AU - Tianyuan Yao
AU - Gelei Xu
AU - John Cannon
AU - Haichun Yang
AU - Yuankai Huo
AU - Mert R. Sabuncu
AU - Ruining Deng
PY - 2026
DA - 2026/09/23
PB - arXiv
DO - 10.48550/arxiv.2609.27208
UR - https://arxiv.org/abs/2609.27208
AB - Spatial transcriptomics (ST) measures gene expression in tissue context, but dense capture grids can be costly and may repeatedly sample morphologically similar regions. Most active learning strategies were developed for categorical labels and independent samples. We conduct a retrospective pool-based benchmark of active learning versus uniform Random sampling for ST, where expression vectors are high-dimensional and continuous and candidates are spatially correlated. Using two fully profiled public ST cohorts, we mask candidate expression vectors and simulate multi-round selection with uncertainty-based Monte Carlo dropout (MC-dropout) and temporal output discrepancy (TOD), and diversity-based CoreSet and TypiClust-inspired selection. We compare 160 completed configurations at 5%, 10%, 30%, and 50% of the fold-wide training spot pool under patient-level cross-validation, with a separate full-label reference. Within each budget, strategies share the selection schedule, morphology-to-expression predictor, and optimization protocol. We assess mean per-gene within-slide Pearson correlation coefficient (PCC), expression-cluster agreement, and Moran's I fidelity. On HER2-positive breast cancer, pooled mean PCC differences from Random across the four active strategies were -0.0176, -0.0117, +0.0056, and +0.0057 at 5%, 10%, 30%, and 50%, respectively. On cutaneous squamous cell carcinoma (cSCC), three strategies were below Random at 5%, and all four were below Random at 10%. On HER2-positive breast cancer, CoreSet and MC-dropout had lower PCC but higher expression-cluster agreement than Random at the two smallest budgets; this pattern did not reproduce on cSCC. Under the reported fixed training horizons, the evaluated active strategies do not consistently improve on Random at small budgets, and rankings depend on the evaluation measure.
ER -