An Augmented High-Dimensional Graphical Lasso Method to Incorporate Prior Biological Knowledge for Global Network Learning

Zhuang, Yonghua; Xing, Fuyong; Ghosh, Debashis; Banaei-Kashani, Farnoush; Bowler, Russell P.; Kechris, Katerina

doi:10.3389/fgene.2021.760299

Cited by 3 publications

(5 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A total of 260 phosphatase genes were retrieved from GeneCards ( https://www.genecards.org/ ), of which only 28 were identified as DEGs with high expression. These 28 DEGs were further analyzed using a Least Absolute Shrinkage and Selection Operator (LASSO)-penalized Cox regression analysis and a univariate Cox analysis of overall survival (OS) to identify phosphatase genes with prognostic values in the TCGA cohort of LumABC [ 30 , 31 ]. The results were visualized in a forest plot and a prognostic model, from which five candidate phosphatase genes were selected for further analysis.…”

Section: Methodsmentioning

confidence: 99%

Autocrine phosphatase PDP2 inhibits ferroptosis by dephosphorylating ACSL4 in the Luminal A Breast Cancer

Zhu,

Huang,

Chen

et al. 2024

PLoS ONE

View full text Add to dashboard Cite

Phosphatases can dephosphorylate phosphorylated kinases, leading to their inactivation, and ferroptosis is a type of cell death. Therefore, our aim is to identify phosphatases associated with ferroptosis by analyzing the differentially expressed genes (DEGs) of the Luminal A Breast Cancer (LumABC) cohort from the Cancer Genome Atlas (TCGA). An analysis of 260 phosphatase genes from the GeneCard database revealed that out of the 28 DEGs with high expression, only the expression of pyruvate dehydrogenase phosphatase 2 (PDP2) had a significant correlation with patient survival. In addition, an analysis of DEGs using gene ontology, Kyoto Encyclopedia of Genes and Genomes and gene set enrichment analysis revealed a significant variation in the expression of ferroptosis-related genes. To further investigate this, we analyzed 34 ferroptosis-related genes from the TCGA-LumABC cohort. The expression of long-chain acyl-CoA synthetase 4 (ACSL4) was found to have the highest correlation with the expression of PDP2, and its expression was also inversely proportional to the survival rate of patients. Western blot experiments using the MCF-7 cell line showed that the phosphorylation level of ACSL4 was significantly lower in cells transfected with the HA-PDP2 plasmid, and ferroptosis was correspondingly reduced (p < 0.001), as indicated by data from flow cytometry detection of membrane-permeability cell death stained with 7-aminoactinomycin, lipid peroxidation, and Fe2+. Immunoprecipitation experiments further revealed that the phosphorylation level of ACSL4 was only significantly reduced in cells where PDP2 and ACSL4 co-precipitated. These findings suggest that PDP2 may act as a phosphatase to dephosphorylate and inhibit the activity of ACSL4, which had been phosphorylated and activated in LumABC cells. Further experiments are needed to confirm the molecular mechanism of PDP2 inhibiting ferroptosis.

show abstract

Section: Methodsmentioning

confidence: 99%

Autocrine phosphatase PDP2 inhibits ferroptosis by dephosphorylating ACSL4 in the Luminal A Breast Cancer

Zhu,

Huang,

Chen

et al. 2024

PLoS ONE

View full text Add to dashboard Cite

show abstract

“…Using the prior PPI network retrieved from the STRING database, we applied our previously developed AhGlasso algorithm [24] on the transcriptomics data of 2656 subjects without proteomics measurements but with known clinical phenotype data to construct COPD-associated networks. In order to find the optimal regularization parameter, λ, networks were estimated under a sequence of λ values.…”

Section: Ppi Reconstruction With Ahglassomentioning

confidence: 99%

“…In order to find the optimal regularization parameter, λ, networks were estimated under a sequence of λ values. The upper bound λ max was calculated as described previously [24]. With predefined 0.01 λ minimal ratio, λ min = 0.01 * λ max .…”

Section: Ppi Reconstruction With Ahglassomentioning

confidence: 99%

“…Then 40 λ candidates were generated between λ min and λ max in a log scale for grid search. The optimal λ was selected using the Bayesian Information Criterion (BIC) based on the penalized log Likelihood with 4-fold cross-validation and one standard error rule [24]. The resultant network serves as the updated COPD-associated PPI network for model development.…”

Section: Ppi Reconstruction With Ahglassomentioning

confidence: 99%

“…However, PPI information specific to COPD in STRING might be limited. To build a more complete and specific protein-protein interaction network related to COPD, we used our previously developed AhGlasso (Augmented High-Dimensional Graphical Lasso Method) method [24] to reconstruct a network, and demonstrated how integrating the updated COPD-associated PPI further improves performance. To evaluate stability of the ConvGNN model, we used cross-validation to measure classification performance.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Deep learning on graphs for multi-omics classification of COPD

et al. 2023

View full text Add to dashboard Cite

Network approaches have successfully been used to help reveal complex mechanisms of diseases including Chronic Obstructive Pulmonary Disease (COPD). However despite recent advances, we remain limited in our ability to incorporate protein-protein interaction (PPI) network information with omics data for disease prediction. New deep learning methods including convolution Graph Neural Network (ConvGNN) has shown great potential for disease classification using transcriptomics data and known PPI networks from existing databases. In this study, we first reconstructed the COPD-associated PPI network through the AhGlasso (Augmented High-Dimensional Graphical Lasso Method) algorithm based on one independent transcriptomics dataset including COPD cases and controls. Then we extended the existing ConvGNN methods to successfully integrate COPD-associated PPI, proteomics, and transcriptomics data and developed a prediction model for COPD classification. This approach improves accuracy over several conventional classification methods and neural networks that do not incorporate network information. We also demonstrated that the updated COPD-associated network developed using AhGlasso further improves prediction accuracy. Although deep neural networks often achieve superior statistical power in classification compared to other methods, it can be very difficult to explain how the model, especially graph neural network(s), makes decisions on the given features and identifies the features that contribute the most to prediction generally and individually. To better explain how the spectral-based Graph Neural Network model(s) works, we applied one unified explainable machine learning method, SHapley Additive exPlanations (SHAP), and identified CXCL11, IL-2, CD48, KIR3DL2, TLR2, BMP10 and several other relevant COPD genes in subnetworks of the ConvGNN model for COPD prediction. Finally, Gene Ontology (GO) enrichment analysis identified glycosaminoglycan, heparin signaling, and carbohydrate derivative signaling pathways significantly enriched in the top important gene/proteins for COPD classifications.

show abstract

DRAGON: Determining Regulatory Associations using Graphical models on multi-Omic Networks

Weighill¹,

Burkholz²,

Guebila³

et al. 2022

Nucleic Acids Research

View full text Add to dashboard Cite

The increasing quantity of multi-omic data, such as methylomic and transcriptomic profiles collected on the same specimen or even on the same cell, provides a unique opportunity to explore the complex interactions that define cell phenotype and govern cellular responses to perturbations. We propose a network approach based on Gaussian Graphical Models (GGMs) that facilitates the joint analysis of paired omics data. This method, called DRAGON (Determining Regulatory Associations using Graphical models on multi-Omic Networks), calibrates its parameters to achieve an optimal trade-off between the network’s complexity and estimation accuracy, while explicitly accounting for the characteristics of each of the assessed omics ‘layers.’ In simulation studies, we show that DRAGON adapts to edge density and feature size differences between omics layers, improving model inference and edge recovery compared to state-of-the-art methods. We further demonstrate in an analysis of joint transcriptome - methylome data from TCGA breast cancer specimens that DRAGON can identify key molecular mechanisms such as gene regulation via promoter methylation. In particular, we identify Transcription Factor AP-2 Beta (TFAP2B) as a potential multi-omic biomarker for basal-type breast cancer. DRAGON is available as open-source code in Python through the Network Zoo package (netZooPy v0.8; netzoo.github.io).

show abstract

An Augmented High-Dimensional Graphical Lasso Method to Incorporate Prior Biological Knowledge for Global Network Learning

Cited by 3 publications

References 39 publications

Autocrine phosphatase PDP2 inhibits ferroptosis by dephosphorylating ACSL4 in the Luminal A Breast Cancer

Autocrine phosphatase PDP2 inhibits ferroptosis by dephosphorylating ACSL4 in the Luminal A Breast Cancer

Deep learning on graphs for multi-omics classification of COPD

DRAGON: Determining Regulatory Associations using Graphical models on multi-Omic Networks

Contact Info

Product

Resources

About