A split-control criterion for few-shot adaptation in single-cell perturbation prediction
- 1Macau University of Science and Technology, Macau, China; Jinan University, Guangzhou, China; Institute for Intelligent Bioinnovation, Knowledge Transfer Center
Abstract
Few-shot adaptation can help or harm single-cell perturbation prediction, but conventional evaluation shares a control estimate between calibration responses and held-out targets. We show how methods carry control error into their predictions and derive its contribution to score inflation. Nested control-depth experiments support inverse-depth scaling of inner-product inflation; correlation inflation also depends on the control noise-to-signal ratio. Independent-control scoring changed the rank of 40 of 52 configuration-dataset entries. We introduce SafeDelta, a split-control criterion that uses calibration data to decide whether to adapt. Across six datasets, mean gated gain was non-negative under the original two-control evaluation. On a deep plate, fixed predictions and decisions retained positive mean gains under three additional independent control references. The same reproducibility principle identified a fibrosis programme that localises to scar in human liver. These results inform adaptation decisions and control-depth planning.
Related
- PerturbBridge: Conditional Latent Schr\"{o}dinger Bridge for Single-Cell Perturbation Prediction
- scDisentangle: sequential disentanglement for single-cell perturbation response prediction
- How many cells resolve a perturbation direction? A closed-form, control-aware cell quota for single-cell perturbation screens
- PopPert: Population-level Joint-Distribution Modeling for Single-Cell Perturbation Prediction
- Single-cell perturbation prediction near experimental reproducibility with GENESIS
BibTeX
@misc{liang2026splitcontrol,
title = {{A split-control criterion for few-shot adaptation in single-cell perturbation prediction}},
author = {W.-L. Liang and Y.-B. Yang and Q.-Y. Zhang and J.-C. Feng and Q.-F. Wang and D. Zou and J.-L. Tang and Y. Liu and S.-Y. Wang and J. Duan and Y. Bai and Y.-F. Li and W.-Y. Sun and Y.-F. Cao and R.-R. He},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.05.756906},
url = {https://www.biorxiv.org/content/10.64898/2026.10.05.756906v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - A split-control criterion for few-shot adaptation in single-cell perturbation prediction
AU - W.-L. Liang
AU - Y.-B. Yang
AU - Q.-Y. Zhang
AU - J.-C. Feng
AU - Q.-F. Wang
AU - D. Zou
AU - J.-L. Tang
AU - Y. Liu
AU - S.-Y. Wang
AU - J. Duan
AU - Y. Bai
AU - Y.-F. Li
AU - W.-Y. Sun
AU - Y.-F. Cao
AU - R.-R. He
PY - 2026
DA - 2026/10/05
PB - bioRxiv
DO - 10.64898/2026.10.05.756906
UR - https://www.biorxiv.org/content/10.64898/2026.10.05.756906v1
AB - Few-shot adaptation can help or harm single-cell perturbation prediction, but conventional evaluation shares a control estimate between calibration responses and held-out targets. We show how methods carry control error into their predictions and derive its contribution to score inflation. Nested control-depth experiments support inverse-depth scaling of inner-product inflation; correlation inflation also depends on the control noise-to-signal ratio. Independent-control scoring changed the rank of 40 of 52 configuration-dataset entries. We introduce SafeDelta, a split-control criterion that uses calibration data to decide whether to adapt. Across six datasets, mean gated gain was non-negative under the original two-control evaluation. On a deep plate, fixed predictions and decisions retained positive mean gains under three additional independent control references. The same reproducibility principle identified a fibrosis programme that localises to scar in human liver. These results inform adaptation decisions and control-depth planning.
ER -