Home Medizin Weiterentwicklung der Genregulationsnetzwerkinferenz durch kausale Entdeckung und grafische neuronale Netzwerke

Weiterentwicklung der Genregulationsnetzwerkinferenz durch kausale Entdeckung und grafische neuronale Netzwerke

von NFI Redaktion

Gene Regulation Networks (GRNs) depict the regulatory mechanisms of genes in cellular systems as a network, providing crucial insights into understanding cell processes and molecular interactions that determine cellular phenotypes. Transcriptional regulation, a predominant type of gene expression regulation, involves controlling target genes (TGs) by transcription factors (TFs). One of the biggest challenges in deriving GRNs is establishing causal relationships and not just correlations between the different components of the system. Therefore, deriving gene regulation networks from the perspective of causality is important in understanding the underlying mechanisms that govern the dynamics of cellular systems.

Recently, Quantitative Biology published an approach titled „Gene Regulatory Network Inference based on Causal Discovery Integrating with Graph Neural Network“, which utilizes learning the graph representation and causal asymmetric learning while accommodating linear as well as non-linear regulatory relationships. GRINCD achieves superior performance in predicting regulatory relationships not only of TF-TG but also of TF-TF, where generalized correlation-based methods are unattainable.

GRINCD uses ensemble learning to predict the causal regulation of each regulator-target pair based on the Additive Noise Model (ANM), using a high-quality representation for each gene generated by the Graph Neural Network as input. Specifically, GRINCD utilizes random walks and degree distributions of the nodes to generate edge annotations, and passes them to a two-layered GraphSAGE connected with a binary classifier to obtain the representation of each node. GRINCD achieves optimal performance across multiple datasets under different evaluation metrics. As an application, GRINCD, by analyzing significant changes in regulatory relationships with the progression of the disease, identifies crucial potential regulators driving the transition from colitis to colon cancer.


Journal Reference:

Feng, K., et al. (2023). Inference of Gene Regulatory Network based on Causal Discovery Integrated with Graph Neural Network. Quantitative Biology. doi.org/10.1002/qub2.26.

Related Posts

Adblock Detected

Please support us by disabling your AdBlocker extension from your browsers for our website.