Saved in:
Bibliographic Details
Main Authors: Panagopoulos, George, Lutzeyer, Johannes F., Ennadir, Sofiane, Vazirgiannis, Michalis, Pang, Jun
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.14571
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908601019269120
author Panagopoulos, George
Lutzeyer, Johannes F.
Ennadir, Sofiane
Vazirgiannis, Michalis
Pang, Jun
author_facet Panagopoulos, George
Lutzeyer, Johannes F.
Ennadir, Sofiane
Vazirgiannis, Michalis
Pang, Jun
contents Genomic studies face a vast hypothesis space, while interventions such as gene perturbations remain costly and time-consuming. To accelerate such experiments, gene perturbation models predict the transcriptional outcome of interventions. Since constructing the training set is challenging, active learning is often employed in a "lab-in-the-loop" process. While this strategy makes training more targeted, it is substantially slower, as it fails to exploit the inherent parallelizability of Perturb-seq experiments. Here, we focus on graph neural network-based gene perturbation models and propose a subset selection method that, unlike active learning, selects the training perturbations in one shot. Our method chooses the interventions that maximize the propagation of the supervision signal to the model. The selection criterion is defined over the input knowledge graph and is optimized with submodular maximization, ensuring a near-optimal guarantee. Experimental results across multiple datasets show that, in addition to providing months of acceleration compared to active learning, the method improves the stability of perturbation choices while maintaining competitive predictive accuracy.
format Preprint
id arxiv_https___arxiv_org_abs_2503_14571
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Efficient Data Selection for Training Genomic Perturbation Models
Panagopoulos, George
Lutzeyer, Johannes F.
Ennadir, Sofiane
Vazirgiannis, Michalis
Pang, Jun
Quantitative Methods
Machine Learning
Genomic studies face a vast hypothesis space, while interventions such as gene perturbations remain costly and time-consuming. To accelerate such experiments, gene perturbation models predict the transcriptional outcome of interventions. Since constructing the training set is challenging, active learning is often employed in a "lab-in-the-loop" process. While this strategy makes training more targeted, it is substantially slower, as it fails to exploit the inherent parallelizability of Perturb-seq experiments. Here, we focus on graph neural network-based gene perturbation models and propose a subset selection method that, unlike active learning, selects the training perturbations in one shot. Our method chooses the interventions that maximize the propagation of the supervision signal to the model. The selection criterion is defined over the input knowledge graph and is optimized with submodular maximization, ensuring a near-optimal guarantee. Experimental results across multiple datasets show that, in addition to providing months of acceleration compared to active learning, the method improves the stability of perturbation choices while maintaining competitive predictive accuracy.
title Efficient Data Selection for Training Genomic Perturbation Models
topic Quantitative Methods
Machine Learning
url https://arxiv.org/abs/2503.14571