Graph Signal Processing, Graph Neural Network and Graph Learning on Biological Data: A Systematic Review

Li, Rui; Yuan, Xin; Radfar, Mohsen; Marendy, Peter; Ni, Wei; O’Brien, Terence J.; Casillas‐Espinosa, Pablo M.

doi:10.1109/rbme.2021.3122522

Cited by 52 publications

(19 citation statements)

References 154 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This section introduces GTV regularization, which is based on the definition of a graph. [33][34][35][36][37] Lu et al discretized the unstructured domain corresponding to complex geometry by an undirected simple graph G = (V, E, W). 21 In the graph , where d 𝑗,k is the Euclidean distance between vertices 𝑗 and k. This means that the closer the two vertices are, the more similar they are.…”

Section: Gtv Regularizationmentioning

confidence: 99%

Diffusion optical tomography reconstruction based on convex–nonconvex graph total variation regularization

Xie

Liu

et al. 2022

Math Methods in App Sciences

View full text Add to dashboard Cite

Graph total variation (GTV) is a powerful regularization tool for diffuse optical tomography (DOT) reconstruction since it combines the powerful representation ability of graph and the edge-preserving ability of total variation (TV) regularization. However, as everyone knows, the classical TV regularization trend underestimates the large edge values. In this paper, we propose a convex-nonconvex graph total variation (CNC-GTV) regularization for DOT reconstruction. In particular, we construct a nonconvex regularization by subtracting the generalized Huber function from the GTV regularization. We show that the global convexity of the objective function can be guaranteed by adjusting the nonconvex control parameters. Moreover, we present an alternating direction multiplier method (ADMM) to solve the proposed DOT reconstruction model. Numerical experiments show that the proposed model outperforms existing models in visual and numerical results.

show abstract

Section: Gtv Regularizationmentioning

confidence: 99%

Diffusion optical tomography reconstruction based on convex–nonconvex graph total variation regularization

Xie

Liu

et al. 2022

Math Methods in App Sciences

View full text Add to dashboard Cite

show abstract

“…In recent years, graph neural networks (GNN) have aroused great attention due to their excellent feature extraction capability for unstructured data and are widely used in the fields of biology [14], medicine [15], transportation [16], and recommender systems [17]. End-to-end microstructure-material property modeling has been done by using graph neural networks.…”

Section: Introductionmentioning

confidence: 99%

A node graph using SEM images for material properties prediction with a case study of composite material

Chen,

Yin,

Song

2024

Phys. Scr.

View full text Add to dashboard Cite

Establishing a mapping model between the microstructure and material properties of composite materials is crucial for material development. Scanning electron microscope (SEM) images are widely used for the prediction of material properties. However, the prediction from a single SEM image is independent and does not fully reflect the microstructure characteristics. To address this issue, this paper proposes a node graph construction strategy for SEM images and establishes a multi-graph-based graph attention network (GAT) material property prediction model to achieve the convergence of mutual complementation in microstructure features by using GAT. Firstly, multiple SEM images are constructed into node graphs by a microstructure feature encoder. Next, the microstructure features of multiple SEM images on the node graphs are mutually complemented and converged by using GAT. Finally, the prediction is carried out by using multiple SEM images. The experimental results show that the proposed method shows better performance than other methods.

show abstract

“…We here address these gaps and propose a new DL-based model for imputation in proteomics datasets that exploits additional proteomics features in the form of amino acid sequences and peptide-protein relationships. As graph neural network (GNN) models have shown considerable success in modeling complex relationships between molecules and learning from biological and omics data [23, 24, 25], our method relies on a GNN architecture. What is more, while most proteomics imputation methods still impute on the protein level, our model acts directly on the peptide level, a strategy shown to yield improved imputation results [9].…”

Section: Introductionmentioning

confidence: 99%

PEPerMINT: Peptide Abundance Imputation in Mass Spectrometry-based Proteomics using Graph Neural Networks

Pietz,

Gupta,

Schlaffner

et al. 2024

Preprint

View full text Add to dashboard Cite

Motivation: Accurate quantitative information about the protein abundance is crucial for understanding a biological system and its dynamics. Protein abundance is commonly estimated using label-free, bottom-up mass spectrometry protocols. Here, proteins are digested into peptides before quantification via mass spectrometry. However, missing peptide abundance values, which can make up more than 50% of all abundance values, are a common issue. They result in missing protein abundance values, which then hinder accurate and reliable downstream analyses. Results: To impute missing abundance values, we propose PEPerMINT, a graph neural network model working directly on the peptide level that flexibly takes both peptide-to-protein relationships in a graph format as well as amino acid sequence information into account. We benchmark our method against eleven common imputation methods on six diverse datasets, including cell lines, tissue, and plasma samples. We observe that PEPerMINT consistently outperforms other imputation methods. Its prediction performance remains high for varying degrees of missingness, different evaluation approaches and differential expression prediction. As an additional novel feature, PEPerMINT provides meaningful uncertainty estimates and allows for tailoring imputation to the user's needs based on the reliability of imputed values. Availability and implementation: The code is available at https://github.com/DILiS-lab/pepermint.

show abstract

Graph Signal Processing, Graph Neural Network and Graph Learning on Biological Data: A Systematic Review

Cited by 52 publications

References 154 publications

Diffusion optical tomography reconstruction based on convex–nonconvex graph total variation regularization

Diffusion optical tomography reconstruction based on convex–nonconvex graph total variation regularization

A node graph using SEM images for material properties prediction with a case study of composite material

PEPerMINT: Peptide Abundance Imputation in Mass Spectrometry-based Proteomics using Graph Neural Networks

Contact Info

Product

Resources

About