English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Motivation: Molecular interaction networks summarize complex biologicalprocesses as graphs, whose structure is informative of biological function atmultiple scales. Simultaneously, omics technologies measure the variation oractivity of genes, proteins, or metabolites across individuals or experimentalconditions. Integrating the complementary viewpoints of biological networks andomics data is an important task in bioinformatics, but existing methods treatnetworks as discrete structures, which are intrinsically difficult to integratewith continuous node features or activity measures. Graph neural networks mapgraph nodes into a low-dimensional vector space representation, and can betrained to preserve both the local graph structure and the similarity betweennode features.  Results: We studied the representation of transcriptional, protein-proteinand genetic interaction networks in E. Coli and mouse using graph neuralnetworks. We found that such representations explain a large proportion ofvariation in gene expression data, and that using gene expression data as nodefeatures improves the reconstruction of the graph from the embedding. Wefurther proposed a new end-to-end graph feature auto-encoder which is trainedon the feature reconstruction task, and showed that it performs better atpredicting unobserved node features than auto-encoders that are trained on thegraph reconstruction task before learning to predict node features. Whenapplied to the problem of imputing missing data in single-cell RNAseq data, ourgraph feature auto-encoder outperformed a state-of-the-art imputation methodthat does not use protein interaction information, showing the benefit ofintegrating biological networks and omics data using graph representationlearning.

A Graph Feature Auto-Encoder for the Prediction of Unobserved Node Features on Biological Networks