Learning Adaptive Perturbation-Conditioned Contexts for Robust Transcriptional Response Prediction

Yinhua Piao, Hyomin Kim, Seonghwan Kim, Yunhak Oh, Junhyeok Jeon, Sang-Yeon Hwang, Jaechang Lim, Woo Youn Kim, Chanyoung Park, Sungsoo Ahn
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:98696-98729, 2026.

Abstract

Predicting high-dimensional transcriptional responses to genetic perturbations is challenging because signals are sparse and experimental noise is severe. Existing methods often suffer from mean collapse, achieving high correlation by predicting the global average expression rather than perturbation-specific responses, which yields false positives and poor interpretability. Methods that add biological knowledge graphs typically treat them as dense, static priors shared across perturbations, propagating noise. We propose AdaPert, which counters mean collapse by extracting a sparse, perturbation-specific subgraph via differentiable node selection, then suppressing spurious variation in non-responsive genes while emphasizing differentially expressed ones. Across multiple benchmarks, AdaPert outperforms existing baselines, with the largest gains on DEG-aware metrics.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-piao26a, title = {Learning Adaptive Perturbation-Conditioned Contexts for Robust Transcriptional Response Prediction}, author = {Piao, Yinhua and Kim, Hyomin and Kim, Seonghwan and Oh, Yunhak and Jeon, Junhyeok and Hwang, Sang-Yeon and Lim, Jaechang and Kim, Woo Youn and Park, Chanyoung and Ahn, Sungsoo}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {98696--98729}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/piao26a/piao26a.pdf}, url = {https://proceedings.mlr.press/v306/piao26a.html}, abstract = {Predicting high-dimensional transcriptional responses to genetic perturbations is challenging because signals are sparse and experimental noise is severe. Existing methods often suffer from mean collapse, achieving high correlation by predicting the global average expression rather than perturbation-specific responses, which yields false positives and poor interpretability. Methods that add biological knowledge graphs typically treat them as dense, static priors shared across perturbations, propagating noise. We propose AdaPert, which counters mean collapse by extracting a sparse, perturbation-specific subgraph via differentiable node selection, then suppressing spurious variation in non-responsive genes while emphasizing differentially expressed ones. Across multiple benchmarks, AdaPert outperforms existing baselines, with the largest gains on DEG-aware metrics.} }
Endnote
%0 Conference Paper %T Learning Adaptive Perturbation-Conditioned Contexts for Robust Transcriptional Response Prediction %A Yinhua Piao %A Hyomin Kim %A Seonghwan Kim %A Yunhak Oh %A Junhyeok Jeon %A Sang-Yeon Hwang %A Jaechang Lim %A Woo Youn Kim %A Chanyoung Park %A Sungsoo Ahn %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-piao26a %I PMLR %P 98696--98729 %U https://proceedings.mlr.press/v306/piao26a.html %V 306 %X Predicting high-dimensional transcriptional responses to genetic perturbations is challenging because signals are sparse and experimental noise is severe. Existing methods often suffer from mean collapse, achieving high correlation by predicting the global average expression rather than perturbation-specific responses, which yields false positives and poor interpretability. Methods that add biological knowledge graphs typically treat them as dense, static priors shared across perturbations, propagating noise. We propose AdaPert, which counters mean collapse by extracting a sparse, perturbation-specific subgraph via differentiable node selection, then suppressing spurious variation in non-responsive genes while emphasizing differentially expressed ones. Across multiple benchmarks, AdaPert outperforms existing baselines, with the largest gains on DEG-aware metrics.
APA
Piao, Y., Kim, H., Kim, S., Oh, Y., Jeon, J., Hwang, S., Lim, J., Kim, W.Y., Park, C. & Ahn, S.. (2026). Learning Adaptive Perturbation-Conditioned Contexts for Robust Transcriptional Response Prediction. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:98696-98729 Available from https://proceedings.mlr.press/v306/piao26a.html.

Related Material