Supervised Graph Contrastive Learning for Gene Regulatory Networks

TL;DR

This paper introduces SupGCL, a supervised graph contrastive learning method that leverages real biological perturbations from gene knockdown experiments to improve gene regulatory network representations, outperforming existing methods.

Contribution

SupGCL is a novel GCL approach that incorporates biological perturbations as supervision, bridging artificial augmentations and real experimental data for enhanced network analysis.

Findings

SupGCL yields clearer disease-subtype structures in embeddings.

It improves clustering and downstream task performance.

Outperforms baseline methods on 13 gene and patient prediction tasks.

Abstract

Graph Contrastive Learning (GCL) is a powerful self-supervised learning framework that performs data augmentation through graph perturbations, with growing applications in the analysis of biological networks such as Gene Regulatory Networks (GRNs). The artificial perturbations commonly used in GCL, such as node dropping, induce structural changes that can diverge from biological reality. This concern has contributed to a broader trend in graph representation learning toward augmentation-free methods, which view such structural changes as problematic and should be avoided. However, this trend overlooks the fundamental insight that structural changes from biologically meaningful perturbations are not a problem to be avoided, but rather a rich source of information, thereby ignoring the valuable opportunity to leverage data from real biological experiments. Motivated by this insight, we propose SupGCL (Supervised Graph Contrastive Learning), a new GCL method for GRNs that directly incorporates biological perturbations from gene knockdown experiments as supervision. SupGCL is a probabilistic formulation that continuously generalizes conventional GCL, linking artificial augmentations with real perturbations measured in knockdown experiments, and using the latter as explicit supervision. On patient-derived GRNs from three cancer types, we train GRN representations with SupGCL and evaluate it in two regimes: (i) embedding space analysis, where it yields clearer disease-subtype structure and improves clustering, and (ii) task-specific fine-tuning, where it consistently outperforms strong graph representation learning baselines on 13 downstream tasks spanning gene-level functional annotation and patient-level prediction.