Zhang J et al. (2021), Developmental gene regulatory network connectio...

ECB-ART-50607

PLoS One 2021 Jan 01;1612:e0261926. doi: 10.1371/journal.pone.0261926.

Show Gene links Show Anatomy links

Developmental gene regulatory network connections predicted by machine learning from gene expression data alone.

Zhang J , Ibrahim F , Najmulski E , Katholos G , Altarawy D , Heath LS , Tulin SL .

???displayArticle.abstract???
Gene regulatory network (GRN) inference can now take advantage of powerful machine learning algorithms to complement traditional experimental methods in building gene networks. However, the dynamical nature of embryonic development-representing the time-dependent interactions between thousands of transcription factors, signaling molecules, and effector genes-is one of the most challenging arenas for GRN prediction. In this work, we show that successful GRN predictions for a developmental network from gene expression data alone can be obtained with the Priors Enriched Absent Knowledge (PEAK) network inference algorithm. PEAK is a noise-robust method that models gene expression dynamics via ordinary differential equations and selects the best network based on information-theoretic criteria coupled with the machine learning algorithm Elastic Net. We test our GRN prediction methodology using two gene expression datasets for the purple sea urchin, Stronglyocentrotus purpuratus, and cross-check our results against existing GRN models that have been constructed and validated by over 30 years of experimental results. Our results find a remarkably high degree of sensitivity in identifying known gene interactions in the network (maximum 81.58%). We also generate novel predictions for interactions that have not yet been described, which provide a resource for researchers to use to further complete the sea urchin GRN. Published ChIPseq data and spatial co-expression analysis further support a subset of the top novel predictions. We conclude that GRN predictions that match known gene interactions can be produced using gene expression data alone from developmental time series experiments.

???displayArticle.pubmedLink??? 34962963
???displayArticle.pmcLink??? PMC8714117
???displayArticle.link??? PLoS One

???attribute.lit??? ???displayArticles.show???

References [+] :

Altarawy, PEAK: Integrating Curated and Noisy Prior Knowledge in Gene Regulatory Network Inference. 2017, Pubmed

	Fig 1. Differentially expressed genes.Intersection of unique differentially expressed genes determined by NOISeq, EdgeR and GFold. The intersection of all 5 methods contains 10,627 unique genes specified as differentially expressed as seen where all the ovals overlap in the center. The number of genes uniquely described as DE by each program is the outermost number closest to the label, 2 for NOISeq (Î»1 = 0.9), 411 for NOISeq (Î»1 = 0.85), 16 for GFold1 (Î»3 = â 1), 0 for GFold2 (Î»4 = â 1.5), and 988 for EdgeR.
	Fig 2. Example PEAK predicted interactions.An interactome visualization of a subset of PEAK-predicted interactions within the top-5 predictions (highest confidence scores) for each gene. Both known interactions (yellow gene boxes and red arrows) and new predicted interactions (blue gene boxes and black arrows) are included among these predictions.
	Fig 3. Half-life value evaluation.Sensitivity as a measure of accuracy for the prediction of ectoderm and endomesoderm gene regulatory relations calculated with 5 different median mRNA half-life settings. For the Transcriptomic RNAseq data, 3hrs, 5hrs, 7hrs, 10hrs, 15hrs were tested. For the Nanostring data, 2hrs, 3hrs, 4hrs, 5hrs, 6hrs was tested.