GRAPE: Heterogeneous Graph Representation Learning for Genetic Perturbation with Coding and Non-Coding Biotype

TL;DR

GRAPE tackles the challenge of predicting genetic perturbation outcomes by integrating multi-modal gene features from descriptions and DNA sequences with a biotype-aware heterogeneous graph neural network. The method initializes features, constructs modality-specific GRNs, refines them with graph structure learning, and performs perturbation prediction via a biotype-aware HGAT, optimized with reconstruction and alignment losses. It demonstrates state-of-the-art performance on public perturbation datasets and shows through ablations that biotype information and GSL substantially enhance representations and predictive accuracy. This approach enables more accurate, biotype-considerate perturbation predictions and opens avenues for zero-shot GRN inference with sufficient data.

Abstract

Predicting genetic perturbations enables the identification of potentially crucial genes prior to wet-lab experiments, significantly improving overall experimental efficiency. Since genes are the foundation of cellular life, building gene regulatory networks (GRN) is essential to understand and predict the effects of genetic perturbations. However, current methods fail to fully leverage gene-related information, and solely rely on simple evaluation metrics to construct coarse-grained GRN. More importantly, they ignore functional differences between biotypes, limiting the ability to capture potential gene interactions. In this work, we leverage pre-trained large language model and DNA sequence model to extract features from gene descriptions and DNA sequence data, respectively, which serve as the initialization for gene representations. Additionally, we introduce gene biotype information for the first time in genetic perturbation, simulating the distinct roles of genes with different biotypes in regulating cellular processes, while capturing implicit gene relationships through graph structure learning (GSL). We propose GRAPE, a heterogeneous graph neural network (HGNN) that leverages gene representations initialized with features from descriptions and sequences, models the distinct roles of genes with different biotypes, and dynamically refines the GRN through GSL. The results on publicly available datasets show that our method achieves state-of-the-art performance.

Figures (5)

• Figure 1: Schematic diagram of genetic perturbation.
• Figure 2: Overview of the GRAPE.
• Figure 3: The UMAP visualization of gene representations. Subfigures A and B show the gene text modality representations and gene sequence modality representations extracted by the pre-trained LLM and DNA sequence model, respectively. Subfigure C illustrates the representations extracted by GRAPE after training.
• Figure 4: The picture shows the distribution of predicted values across samples.
• Figure 5: Ablation Study Results.