Graph representation learning in biomedicine and healthcare

In the paper

“Alternatively, variational graph autoencoders produce cell embeddings and interpretable attention weights indicating what genes the model attends to when deriving an embedding for a given cell [159].”

What the sources say

• Partly supported

  Section 4.1 describes graph autoencoders that produce node embeddings and attention weights for different similarity views. The Introduction identifies the nodes as drugs, not cells; the attention weights concern drug-similarity views, not genes. The paper does not describe its model as variational. Page numbers were not available in our copy. (LLM, conf=0.88)

  cites [159] Ma, T., Xiao, C., Zhou, J. & Wang, F. Drug similarity integration through attentive multi-view graph auto-encoders. In Proc. Twenty-Seventh International Joint Conference on Artificial Intelligence (ed. Lang, J.) 3477–3483 (AAAI, 2018). sourceExperimental — accuracy not yet measured

In the paper

“Cell similarity graphs have also been created using autoencoders by first embedding gene expression readouts and then connecting genes based on how similar their embeddings are [157].”

What the sources say

• Partly supported

  The Methods section, “Graph embedding for feature learning” (page unavailable), describes learning node2vec embeddings and using cosine similarity to add edges between drugs or between targets. This supports the general embedding-then-connect-by-similarity approach, but not the attributed use of autoencoders, gene-expression readouts, or cell similarity graphs. (LLM, conf=0.84)

  cites [157] Thafar, M. A. et al. DTiGEMS+: drug–target interaction prediction using graph embedding, graph mining, and similarity-based techniques. J. Cheminformatics 12, 44 (2020). sourceExperimental — accuracy not yet measured

In the paper

“In fact, as ligand-receptor interactions are directed, they could be used to infer causal interactions of previously unknown ligand-receptor pairs [161, 163].”

What the sources say

• Partly supported

  The Introduction describes predicting compound–protein interactions, and §2.2.2 describes predicting whether particular ligand–protein pairs interact. Neither passage says that ligand–receptor interactions are directed or that their direction can be used to infer causal interactions; that is the unsupported part of the claim. Page numbers are unavailable in our copy. (LLM, conf=0.84)

  cites [161] Chen, L. et al. TransformerCPI: improving compound–protein interaction prediction by sequence-based deep learning with self-attention mechanism and label reversal experiments. Bioinformatics 36, 4406–4414 (2020). sourceExperimental — accuracy not yet measured
• Partly supported

  The Introduction (page unavailable) says that identifying previously unknown compound–protein interactions is useful, and describes methods for predicting them. It does not say that ligand–receptor interactions are directed or that their direction can be used to infer causal interactions. (LLM, conf=0.83)

  cites [163] Tsubaki, M., Tomii, K. & Sese, J. Compound–protein interaction prediction with end-to-end learning of neural networks for graphs and sequences. Bioinformatics 35, 309–318 (2019). sourceExperimental — accuracy not yet measured

In the paper

“Two modes of patient data successfully integrated using deep graph learning are histopathological images [8, 185] and EHRs [186, 187].”

What the sources say

• Supported

  This source is cited for histopathological images, not EHRs, which have separate citations. In Methods, the paper describes constructing cell graphs from histopathology images and using graph convolutional networks to learn representations for outcome prediction; it also uses those representations in Pathomic Fusion. Page numbers are not available in our copy. (LLM, conf=0.90)

  cites [8] Chen, R. J. et al. Pathomic fusion: an integrated framework for fusing histopathology and genomic features for cancer diagnosis and prognosis. In IEEE Transactions on Medical Imaging Vol. 41, 757–770 (IEEE, 2022). source
• Not checked yetcites [185] Levy, J., Haudenschild, C., Bar, C., Christensen, B. & Vaickus, L. Topological feature extraction and visualization of whole slide images using graph neural networks. Pac. Symp. Biocomput. 26, 285–296 (2021).
• Supported

  The citation to this paper is attached to histopathological images, not EHRs. In “Overview of SpaGCN and evaluation,” the paper says SpaGCN uses histology information to build a graph and a graph convolutional layer to analyze gene expression. “Application to human primary pancreatic cancer tissue” reports that its detected domain corresponded better to the cancer region annotated on the histology image. (LLM, conf=0.91)

  cites [186] Hu, J. et al. SpaGCN: integrating gene expression, spatial location and histology to identify spatial domains and spatially variable genes by graph convolutional network. Nat. Methods 18, 1342–1351 (2021). source
• Partly supported

  The Contributions section reports learning a disease–symptom knowledge graph from electronic medical records, supporting the EHR portion of the cited sentence. But Methods > Algorithms for constructing a knowledge graph specifies logistic regression, naive Bayes, and a noisy-OR Bayesian network—not deep graph learning. The histopathological-image portion is cited to other sources and is not assessed here. (LLM, conf=0.89)

  cites [187] Rotmensch, M., Halpern, Y., Tlimat, A., Horng, S. & Sontag, D. Learning a health knowledge graph from electronic medical records. Sci. Rep. 7, 5994 (2017). sourceExperimental — accuracy not yet measured

In the paper

“EHRs also have underlying spatial and/or temporal dependencies [206] that many methods have recently taken advantage of to perform time-dependent prediction tasks.”

What the sources say

• ?Could not verify

  The Introduction and Section 2.2 discuss spatial dependencies between DNA regions, not electronic health records. Searches for EHRs, electronic health records, patient records, and time-dependent prediction found no passage addressing the attributed claim. (LLM, conf=0.78)

  cites [206] Lanchantin, J. & Qi, Y. Graph convolutional networks for epigenetic state prediction using both sequence and 3D genome data. Bioinformatics 36, i659–i667 (2020). sourceExperimental — accuracy not yet measured

In the paper

“In contrast to shallow embeddings, GNNs can capture higher-order and non-linear patterns through multi-hop propagation within several layers of neural message passing.”

In the paper

“Given by the shared components and disease module hypotheses [13], cellular components associated with the same phenotype tend to cluster in the same network neighborhood [16].”

What the sources say

• Not checked yetcites [13] Barabási, A.-L., Gulbahce, N. & Loscalzo, J. Network medicine: a network-based approach to human disease. Nat. Rev. Genet. 12, 56–68 (2011). source
• Not checked yetcites [16] Agrawal, M., Zitnik, M. & Leskovec, J. Large-scale analysis of disease pathways in the human interactome. In Pac. Symp. Biocomput. 23, 111–122 (2018).

In the paper

“Topology-based methods have been shown to capture and leverage the dynamics of biological systems to enrich existing PPI networks [119].”

What the sources say

• Not checked yetcites [119] Moreau, Y. & Tranchevent, L.-C. Computational tools for prioritizing candidate genes: boosting disease gene discovery. Nat. Rev. Genet. 13, 523–536 (2012). source

In the paper

“Specifically, the inductive ability of graph convolution neural networks to generalize to data points unseen during model training (Section 2.2) and even generate new data points from scratch by decoding latent representation from the embedding space (Section 2.3) has enabled the discovery of new molecules, interactions, and functions [101, 108, 109].”

What the sources say

• Not checked yetcites [101] Huang, K., Xiao, C., Glass, L. M., Zitnik, M. & Sun, J. SkipGNN: predicting molecular interactions with skip-graph networks. Sci. Rep. 10, 21092 (2020). source
• Not checked yetcites [108] Fout, A., Byrd, J., Shariat, B. & Ben-Hur, A. Protein interface prediction using graph convolutional networks. In Proc. International Conference on Neural Information Processing Systems (eds Guyon, I. et al.) (Curran Associates, 2019).
• Not checked yetcites [109] Ingraham, J., Garg, V. K., Barzilay, R. & Jaakkola, T. S. Generative models for graph-based protein design. In Proc. International Conference on Neural Information Processing Systems (eds Wallach, H. et al.) (Curran Associates, 2019).

In the paper

“node embeddings can capture whether the nodes lie within a hub based on their degree, an important aspect of local neighborhood (e.g., strongly clustered gene embeddings indicate essential housekeeping roles)”

In the paper

“nodes in the same positional regions should have similar embeddings due to the local hypothesis (e.g., highly similar pairs of protein embeddings suggest similar phenotypic consequence)”

In the paper

“Learned embeddings—using graph representation learning methods—that capture genome-wide interactions has enhanced disease predictions, even at the resolution of tissues and single cells.”

What the sources say

• Not checked yetcites [16] Agrawal, M., Zitnik, M. & Leskovec, J. Large-scale analysis of disease pathways in the human interactome. In Pac. Symp. Biocomput. 23, 111–122 (2018).
• Not checked yetcites [45] Menche, J. et al. Uncovering disease-disease relationships through the incomplete interactome. Science 347, 1257601 (2015). source

In the paper

“In protein interaction networks, for instance, mutations in interacting proteins often lead to similar diseases [13].”

What the sources say

• Not checked yetcites [13] Barabási, A.-L., Gulbahce, N. & Loscalzo, J. Network medicine: a network-based approach to human disease. Nat. Rev. Genet. 12, 56–68 (2011). source

In the paper

“We posit that representation learning can realize network biomedicine principles.”

In the paper

“Generative graph models generate new node and edge structures—even entire graphs—that are likely to have desired properties, such as novel molecules with acceptable toxicity profiles (Figure 2).”

In the paper

“Many have integrated various data modalities, including chemical structure, binding affinities, physical and chemical principles, and amino acid sequences, to improve protein interaction quantification [33].”

What the sources say

• Not checked yetcites [33] David, L., Thakkar, A., Mercado, R. & Engkvist, O. Molecular representations in AI-driven drug discovery: a review and practical guide. J. Cheminformatics 12, 56 (2020). source

In the paper

“given by the shared components hypothesis, two nodes with significantly overlapping sets of network neighbors should have similar embeddings due to shared message passing (e.g., highly similar disease embeddings imply shared disease-associated cellular components).”

In the paper

“Characterizing a protein’s function in specific biological contexts is a challenging and experimentally intensive task [122, 123].”

What the sources say

• Not checked yetcites [122] The Gene Ontology Consortium. The Gene Ontology resource: 20 years and still GOing strong. Nucleic Acids Res 47, D330–D338 (2019). source
• Not checked yetcites [123] Zhou, G., Wang, J., Zhang, X. & Yu, G. DeepGOA: predicting gene ontology annotations of proteins via graph convolutional network. In IEEE International Conference on Bioinformatics and Biomedicine (BIBM) 1836–1841 (IEEE, 2019). source

In the paper

“Unlike machine learning approaches designed to analyze data modalities like medical images and biological sequences, graph representation learning methods are uniquely able to leverage structural information in multimodal datasets [101].”

What the sources say

• Not checked yetcites [101] Huang, K., Xiao, C., Glass, L. M., Zitnik, M. & Sun, J. SkipGNN: predicting molecular interactions with skip-graph networks. Sci. Rep. 10, 21092 (2020). source

In the paper

“As gene expression is the direct readout of perturbation effects, changes in gene expression are often used to model disease-specific co-expression or regulatory interactions between genes.”

In the paper

“Due to experimental and resource constraints, the most updated PPI networks are still limited in their number of nodes (proteins) and edges (physical interactions) [118].”

What the sources say

• Not checked yetcites [118] Yao, H., Guan, J. & Liu, T. Denoising protein-protein interaction network via variational graph auto-encoder for protein complex detection. J. Bioinform. Comput. Biol. 18, 2040010 (2020). source

In the paper

“First, interacting entities are typically more similar than non-interacting entities, as implicated by the local hypothesis [13].”

What the sources say

• Not checked yetcites [13] Barabási, A.-L., Gulbahce, N. & Loscalzo, J. Network medicine: a network-based approach to human disease. Nat. Rev. Genet. 12, 56–68 (2011). source

In the paper

“Classic deep learning methods are unable to consider such diverse structural properties and rich interactions, which are the essence of biomedical networks, because classic deep methods are designed for fixed-size grids (e.g., images and tabular datasets) or optimized for text and sequences.”

In the paper

“Grounded in network medicine principles, graph machine learning is commonly applied to learn molecular representations of proteins and their physical interactions for predicting protein function.”

In the paper

“Methods that rely solely on gene expression data typically transform the co-expression matrix into a more topologically meaningful form [146, 147, 148].”

What the sources say

• Not checked yetcites [146] Buterez, D., Bica, I., Tariq, I., Andrés-Terré, H. & Liò, P. CellVGAE: an unsupervised scRNA-seq analysis workflow with graph attention networks. Bioinformatics 38, 1277–1286 (2021). source
• Not checked yetcites [147] Marx, V. Method of the year: spatially resolved transcriptomics. Nat. Methods 18, 9–14 (2021). source
• Not checked yetcites [148] Yuan, Y. & Bar-Joseph, Z. GCNG: graph convolutional networks for inferring gene interaction from spatial transcriptomics data. Genome Biol 21, 300 (2020). source

In the paper

“Thus, the network parsimony principle dictates that shortest molecular paths between known disease-associated components tend to correlate with causal molecular pathways [13].”

What the sources say

• Not checked yetcites [13] Barabási, A.-L., Gulbahce, N. & Loscalzo, J. Network medicine: a network-based approach to human disease. Nat. Rev. Genet. 12, 56–68 (2011). source

In the paper

“The organizing principles governing networks link network structure to molecular phenotypes, biological roles, disease, and health, thus providing the conceptual grounding that, we argue, can explain the successes (and limitations) of graph representation learning and inform future development of the field.”

What the sources say

• Not checked yetcites [12] Goh, K.-I. et al. The human disease network. Proc. Natl Acad. Sci. USA 104, 8685–8690 (2007). source
• Not checked yetcites [13] Barabási, A.-L., Gulbahce, N. & Loscalzo, J. Network medicine: a network-based approach to human disease. Nat. Rev. Genet. 12, 56–68 (2011). source
• Not checked yetcites [14] Hu, J. X., Thomas, C. E. & Brunak, S. Network biology concepts in complex disease comorbidities. Nat. Rev. Genet. 17, 615–629 (2016). source
• Not checked yetcites [15] Zitnik, M. et al. Evolution of resilience in protein interactomes across the tree of life. Proc. Natl Acad. Sci. USA 116, 4426–4433 (2019). source

In the paper

“Graph representation learning methods have been successful in integrating patient records with molecular, genomic, and disease networks to personalize patient predictions.”

What the sources say

• Not checked yetcites [104] Kearnes, S., McCloskey, K., Berndl, M., Pande, V. & Riley, P. Molecular graph convolutions: moving beyond fingerprints. J. Comput. Aided Mol. Des. 30, 595–608 (2016). source
• Not checked yetcites [105] Dutil, F., Cohen, J. P., Weiss, M., Derevyanko, G. & Bengio, Y. Towards gene expression convolutions using gene interaction graphs. In International Conference on Machine Learning Workshop on Computational Biology (2018).

In the paper

“Because gene expression data can be noisy and variational, recent advances include fusing the co-expression matrices with existing biomedical networks, such as GO annotations and PPI, and feeding the resulting graph into graph convolution layers (Section 2.2) [150, 151, 152].”

What the sources say

• Not checked yetcites [150] Meinshausen, N. et al. Methods for causal inference from gene perturbation experiments and validation. Proc. Natl Acad. Sci. USA 113, 7361–7368 (2016). source
• Not checked yetcites [151] Guney, E., Menche, J., Vidal, M. & Barábasi, A.-L. Network-based in silico drug efficacy screening. Nat. Commun. 7, 10331 (2016). source
• Not checked yetcites [152] Cheng, F. et al. Network-based approach to prediction and population-based validation of in silico drug repurposing. Nat. Commun. 9, 2691 (2018). source

In the paper

“Because such processes are costly and time-consuming, in silico approaches have been adopted into the drug discovery pipeline.”

What the sources say

• Not checked yetcites [13] Barabási, A.-L., Gulbahce, N. & Loscalzo, J. Network medicine: a network-based approach to human disease. Nat. Rev. Genet. 12, 56–68 (2011). source
• Not checked yetcites [14] Hu, J. X., Thomas, C. E. & Brunak, S. Network biology concepts in complex disease comorbidities. Nat. Rev. Genet. 17, 615–629 (2016). source
• Not checked yetcites [32] Rai, A., Shinde, P. & Jalan, S. Network spectra for drug-target identification in complex diseases: new guns against old foes. Appl. Netw. Sci. 3, 51 (2018). source
• Not checked yetcites [164] Lin, X. et al. DeepGS: Deep representation learning of graphs and sequences for drug-target binding affinity prediction. ECAI (2020).

In the paper

“Graph representation learning has been successful in characterizing drugs at the systems level without patient data to make predictions about interactions with other drugs, protein targets, side effects, diseases [6, 38, 39, 40, 48, 165].”

What the sources say

• Not checked yetcites [6] Gysi, D. M. et al. Network medicine framework for identifying drug-repurposing opportunities for COVID-19. Proc. Natl Acad. Sci. USA 118, e2025581118 (2021). source
• Not checked yetcites [38] Gaudelet, T. et al. Utilizing graph machine learning within drug discovery and development. Brief. Bioinform. 22, bbab159 (2021). source
• Not checked yetcites [39] MacLean, F. Knowledge graphs and their applications in drug discovery. Expert Opin. Drug Discov 16, 1057–1069 (2021). source
• Not checked yetcites [40] Zeng, X., Tu, X., Liu, Y., Fu, X. & Su, Y. Toward better drug discovery with knowledge graph. Curr. Opin. Struct. Biol. 72, 114–126 (2022). source
• Not checked yetcites [48] Cheng, F. et al. A genome-wide positioning systems network algorithm for in silico drug repurposing. Nat. Commun. 10, 3476 (2019). source
• Not checked yetcites [165] Guo, Z.-H. et al. MeSHHeading2vec: a new method for representing mesh headings as vectors based on graph embedding algorithm. Brief. Bioinform. 21, 1641–1662 (2020). source

In the paper

“Single-cell RNA sequencing (scRNA-seq) data lend themselves to graph representation learning to model cellular differential processes [136, 155] and disease states [140].”

What the sources say

• Not checked yetcites [136] Crawford, J. & Greene, C. S. Incorporating biological structure into machine learning models in biomedicine. Curr. Opin. Biotechnol. 63, 126–134 (2020). source
• Not checked yetcites [155] Xie, Y. et al. MARS: Markov molecular sampling for multi-objective drug discovery. In International Conference on Learning Representations (2021).
• Not checked yetcites [140] Rizvi, A. H. et al. Single-cell topological rna-seq analysis reveals insights into cellular differentiation and development. Nat. Biotechnol. 35, 551–560 (2017). source

In the paper

“Topological landscapes present in gene expression data can be vectorized and fed into a GCN to classify the disease type [147].”

What the sources say

• Not checked yetcites [147] Marx, V. Method of the year: spatially resolved transcriptomics. Nat. Methods 18, 9–14 (2021). source

In the paper

“Alternatively, gene expression data can directly be used to construct disease and gene networks that are then input into a joint matrix factorization and GCN method to draw disease-gene associations, akin to a recommendation task (Section 2.2) [146].”

What the sources say

• Not checked yetcites [146] Buterez, D., Bica, I., Tariq, I., Andrés-Terré, H. & Liò, P. CellVGAE: an unsupervised scRNA-seq analysis workflow with graph attention networks. Bioinformatics 38, 1277–1286 (2021). source

In the paper

“Further, applying a joint GCN, VAE, and GAN framework (Section 2.3) to gene correlation networks—initialized with a subset of gene expression matrices—can generate disease networks with the desired properties [149].”

What the sources say

• Not checked yetcites [149] Partel, G. & Wählby, C. Spage2vec: Unsupervised representation of localized spatial gene expression signatures. FEBS J 288, 1859–1870 (2021). source

In the paper

“Beyond GNNs and graph autoencoders, learning a manifold over a cell state space can quantify the effects of experimental perturbations [155].”

What the sources say

• Not checked yetcites [155] Xie, Y. et al. MARS: Markov molecular sampling for multi-objective drug discovery. In International Conference on Learning Representations (2021).

In the paper

“However, despite the utility of PPI networks, they have been reported to limit models trained solely on PPI networks because they are unable to capture all gene regulatory activities [153].”

What the sources say

• Not checked yetcites [153] Stokes, J. M. et al. A deep learning approach to antibiotic discovery. Cell 180, 688–702.e13 (2020). source

In the paper

“Gene expression data can be transformed into a colored graph that captures the shape of the data (e.g., using TDA [148]; refer to SI Note 3), which then enables downstream analysis using network science metrics and graph machine learning.”

What the sources say

• Not checked yetcites [148] Yuan, Y. & Bar-Joseph, Z. GCNG: graph convolutional networks for inferring gene interaction from spatial transcriptomics data. Genome Biol 21, 300 (2020). source

In the paper

“Doing so has enabled more interpretable disease classification models (e.g., weighting gene interactions based on existing biological knowledge).”

What the sources say

• Not checked yetcites [150] Meinshausen, N. et al. Methods for causal inference from gene perturbation experiments and validation. Proc. Natl Acad. Sci. USA 113, 7361–7368 (2016). source
• Not checked yetcites [151] Guney, E., Menche, J., Vidal, M. & Barábasi, A.-L. Network-based in silico drug efficacy screening. Nat. Commun. 7, 10331 (2016). source
• Not checked yetcites [152] Cheng, F. et al. Network-based approach to prediction and population-based validation of in silico drug repurposing. Nat. Commun. 9, 2691 (2018). source

In the paper

“Creating cell neighborhood and spatial gene expression graphs require a distance metric, as edges are determined based on spatial proximity, while gene co-expression and molecular similarity graphs need a threshold applied on the gene expression data [35].”

What the sources say

• Not checked yetcites [35] Hetzel, L., Fischer, D. S., Günnemann, S. & Theis, F. J. Graph representation learning for single cell biology. Curr. Opin. Syst. Biol. 28, 100347 (2021). source

In the paper

“For instance, a cell neighborhood graph and a gene pair expression matrix enable classic GNNs to predict ligand-receptor interactions [161].”

What the sources say

• Not checked yetcites [161] Chen, L. et al. TransformerCPI: improving compound–protein interaction prediction by sequence-based deep learning with self-attention mechanism and label reversal experiments. Bioinformatics 36, 4406–4414 (2020). source

In the paper

“As a result, approaches that integrate cross-domain knowledge as a vast heterogeneous network and/or into the model’s architecture seem better equipped to elucidate drug action.”

In the paper

“Fusing compound sequence, structure, and clinical implications has significantly improved drug-drug and drug-target interaction predictions.”

What the sources say

• Not checked yetcites [172] Gurcan, M. N. et al. Histopathological image analysis: a review. IEEE Rev. Biomed. Eng. 2, 147–171 (2009). source
• Not checked yetcites [173] Choi, E., Bahadori, M. T., Song, L., Stewart, W. F. & Sun, J. GRAM: graph-based attention model for healthcare representation learning. In Proc. 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining 787–795 (ACM, 2017). source
• Not checked yetcites [174] Li, Y., Qian, B., Zhang, X. & Liu, H. Graph neural network-based diagnosis prediction. Big Data 8, 379–390 (2020). source
• Not checked yetcites [175] Bonito, P. et al. in Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, and Graphs in Biomedical Image Analysis (eds Sudre, C. H. et al.) 208–219 (Springer, 2020).
• Not checked yetcites [176] Adnan, M., Kalra, S. & Tizhoosh, H. R. Representation learning of histopathology images using graph neural networks. In Proc. IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops 4254–4261 (IEEE, 2020). source
• Not checked yetcites [177] Anand, D., Gadiya, S. & Sethi, A. Histographs: graphs in histopathology. In Medical Imaging 2020: Digital Pathology (eds Tomaszewski, J. E. & Ward, A. D.) (SPIE, 2020). source

In the paper

“But, in addition to high financial cost, the experiments required to measure drug-drug interactions and toxicity face a combinatorial explosion problem [33].”

What the sources say

• Not checked yetcites [33] David, L., Thakkar, A., Mercado, R. & Engkvist, O. Molecular representations in AI-driven drug discovery: a review and practical guide. J. Cheminformatics 12, 56 (2020). source

In the paper

“An example aggregation method is pooling with an attention mechanism to infer relevant patches in the image [190].”

What the sources say

• Not checked yetcites [190] Ma, F. et al. KAME: knowledge-based attention model for diagnosis prediction in healthcare. In 27th ACM International Conference on Information and Knowledge Management 743–752 (ACM, 2018). source

In the paper

“To be flexible with other graph- and non-graph-based methods, protein structure representations generated by graph convolutions has been combined with protein sequence representations (e.g., shallow network embedding methods or CNNs) to predict the probability of compound-protein interactions [174, 175, 176, 177].”

What the sources say

• Not checked yetcites [174] Li, Y., Qian, B., Zhang, X. & Liu, H. Graph neural network-based diagnosis prediction. Big Data 8, 379–390 (2020). source
• Not checked yetcites [175] Bonito, P. et al. in Uncertainty for Safe Utilization of Machine Learning in Medical Imaging, and Graphs in Biomedical Image Analysis (eds Sudre, C. H. et al.) 208–219 (Springer, 2020).
• Not checked yetcites [176] Adnan, M., Kalra, S. & Tizhoosh, H. R. Representation learning of histopathology images using graph neural networks. In Proc. IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops 4254–4261 (IEEE, 2020). source
• Not checked yetcites [177] Anand, D., Gadiya, S. & Sethi, A. Histographs: graphs in histopathology. In Medical Imaging 2020: Digital Pathology (eds Tomaszewski, J. E. & Ward, A. D.) (SPIE, 2020). source

In the paper

“For example, attention mechanisms have been applied on drug graphs, with chemical structures and side effects as features, to generate interpretable predictions of drug-drug interactions [172].”

What the sources say

• Not checked yetcites [172] Gurcan, M. N. et al. Histopathological image analysis: a review. IEEE Rev. Biomed. Eng. 2, 147–171 (2009). source

In the paper

“For example, generating representations of the atoms, distances, and angles to be propagated along the molecular graph has allowed us to identify the angle and direction of interactions between atoms [84].”

What the sources say

• Not checked yetcites [84] Wang, X. et al. Heterogeneous graph attention network. In Proc. ACM Web Conference (eds Liu, L. & White, R.) 2022–2032 (ACM, 2019). source

In the paper

“TDA (refer to SI Note 3) has also been applied to construct graphs of drugs, targets, and diseases separately, from which representations of such entities are learned and optimized for downstream prediction [179].”

What the sources say

• Not checked yetcites [179] Jaume, G. et al. Towards explainable graph representations in digital pathology. In International Conference on Machine Learning Workshop on Computational Biology (2020).

In the paper

“Cell morphology and tissue micro-architecture information can be aggregated from cell graphs to grade cancer histology images (e.g., using GNNs (Section 2.2)) [8, 190, 191, 192].”

What the sources say

• Not checked yetcites [8] Chen, R. J. et al. Pathomic fusion: an integrated framework for fusing histopathology and genomic features for cancer diagnosis and prognosis. In IEEE Transactions on Medical Imaging Vol. 41, 757–770 (IEEE, 2022). source
• Not checked yetcites [190] Ma, F. et al. KAME: knowledge-based attention model for diagnosis prediction in healthcare. In 27th ACM International Conference on Information and Knowledge Management 743–752 (ACM, 2018). source
• Not checked yetcites [191] Sun, Z. et al. Disease prediction via graph neural networks. IEEE J. Biomed. Health Inform. 25, 818–826 (2020). source
• Not checked yetcites [192] Chen, I. Y., Agrawal, M., Horng, S. & Sontag, D. Robustly extracting medical knowledge from EHRs: a case study of learning a health knowledge graph. In Pac. Symp. Biocomput. 25, 19–30 (2020).

In the paper

“Shallow network embedding techniques are also used to generate embeddings for drugs and targets by computing drug-drug, drug-target, and target-target similarities (Section 2.1) [170].”

What the sources say

• Not checked yetcites [170] Buphamalai, P., Kokotovic, T., Nagy, V. & Menche, J. Network analysis reveals rare disease signatures across multiple levels of biological organization. Nat. Commun. 12, 6306 (2021). source

In the paper

“Representing molecules as graphs has improved predictions on various quantum chemistry properties.”

What the sources say

• Not checked yetcites [166] Wang, R., Li, S., Cheng, L., Wong, M. H. & Leung, K. S. Predicting associations among drugs, targets and diseases by tensor decomposition for drug repositioning. BMC Bioinf. 20, 628 (2019). source

In the paper

“Because interpretability is critical for models aimed to generate patient-specific predictions, post-hoc graph pruning optimization may be performed on a cell graph generated from a histopathology image to define subgraphs explaining the original cell graph analysis [193].”

What the sources say

• Not checked yetcites [193] Chowdhury, S., Zhang, C., Yu, P. S. & Luo, Y. Mixed pooling multi-view attention autoencoder for representation learning in healthcare. Preprint at https://arxiv.org/abs/1910.06456 (2019).

In the paper

“However, quantifying these metrics requires labor- and cost-intensive experiments [33, 34].”

What the sources say

• Not checked yetcites [33] David, L., Thakkar, A., Mercado, R. & Engkvist, O. Molecular representations in AI-driven drug discovery: a review and practical guide. J. Cheminformatics 12, 56 (2020). source
• Not checked yetcites [34] Wieder, O. et al. A compact review of molecular property prediction with graph neural networks. Drug Discov. Today. Technol. 37, 1–12 (2020). source

In the paper

“In other instances, PPI networks are combined with genomic features to predict drug sensitivity using GNNs [181].”

What the sources say

• Not checked yetcites [181] An, X., Zhou, Y., Di, Y. & Ming, D. Dynamic functional connectivity and graph convolution network for Alzheimer’s disease classification. In 7th International Conference on Biomedical and Bioinformatics Engineering 1–4 (ACM, 2020). source

In the paper

“Producing atom-centered representations based on a weighted combination of their neighbors’ features (i.e., using an attention mechanism) is able to model interactions among reactants for predicting organic reaction outcomes [167].”

What the sources say

• Not checked yetcites [167] Wan, F., Hong, L., Xiao, A., Jiang, T. & Zeng, J. NeoDTI: neural integration of neighbor information from a heterogeneous network for discovering new drug–target interactions. Bioinformatics 35, 104–111 (2019). source

In the paper

“Modeling small compounds’ and protein targets’ molecular structure as well as their binding affinities and specificity, for instance, using graph representation learning has enabled accelerated investigation of interactions between a given drug and protein target.”

What the sources say

• Not checked yetcites [169] Alsentzer, E., Finlayson, S. G., Li, M. M. & Zitnik, M. Subgraph neural networks. In Proc. International Conference on Neural Information Processing Systems (eds Larochelle, H. et al.) 8017–8029 (Curran Associates, 2020).
• Not checked yetcites [170] Buphamalai, P., Kokotovic, T., Nagy, V. & Menche, J. Network analysis reveals rare disease signatures across multiple levels of biological organization. Nat. Commun. 12, 6306 (2021). source
• Not checked yetcites [171] Barisoni, L., Lafata, K. J., Hewitt, S. M., Madabhushi, A. & Balis, U. G. Digital pathology and computational image analysis in nephropathology. Nat. Rev. Nephrol. 16, 669–685 (2020). source

In the paper

“Due to the major challenge of finding novel and diverse compounds with specific chemical properties, iteratively editing fragments of a molecular graph during training has improved predictions for high-quality candidates targeting our proteins of interest [168].”

What the sources say

• Not checked yetcites [168] Xie, Y., Peng, J. & Zhou, Y. Integrating protein–protein interaction information into drug response prediction by graph neural encoding. Preprint at Research Square https://doi.org/10.21203/rs.2.18936/v1 (2019). source

In the paper

“Alternatively, molecular graphs have been decomposed into a “junction tree,” where each node represents a substructure in the molecule, to learn representations of both the molecular graph and the junction tree for generating new molecules with desirable properties (Section 2.3) [64].”

What the sources say

• Not checked yetcites [64] Yang, B., Yih, W., He, X., Gao, J. & Deng, L. Embedding entities and relations for learning and inference in knowledge bases. In International Conference on Learning Representations (2015).

In the paper

“Deep graph learning has been shown to detect subtle signs of disease progression in the images while integrating other modalities (e.g., tissue localization [189] and genomic features [8]) to improve medical image processing.”

What the sources say

• Not checked yetcites [189] Mao, C., Yao, L. & Luo, Y. MedGCN: medication recommendation and lab test imputation via graph convolutional networks. J. Biomed. Inform. 127, 104000 (2022). source
• Not checked yetcites [8] Chen, R. J. et al. Pathomic fusion: an integrated framework for fusing histopathology and genomic features for cancer diagnosis and prognosis. In IEEE Transactions on Medical Imaging Vol. 41, 757–770 (IEEE, 2022). source

In the paper

“For example, neighborhood information are aggregated from heterogeneous networks comprised of drug, target, and disease information to predict drug-target interactions (Section 2.2) [180].”

What the sources say

• Not checked yetcites [180] Chao, C.-H. et al. Lymph node gross tumor volume detection in oncology imaging via relationship learning using graph neural network. In 23rd International Conference on Medical Image Computing and Computer Assisted Intervention (eds Martel, A. L. et al.) 772–782 (Springer-Verlag, 2020). source

In the paper

“MRI images can be converted into graphs that GNNs are applied to for classifying the progression of Alzheimer’s Disease [195, 196, 197].”

What the sources say

• Not checked yetcites [195] Lee, D., Jiang, X. & Yu, H. Harmonized representation learning on dynamic her graphs. J. Biomed. Inform. 106, 103426 (2020). source
• Not checked yetcites [196] Tong, C., Rocheteau, E., Veličković, P., Lane, N. & Liò, P. in AI for DiseaseSurveillance and Pandemic Intelligence. W3PHAI 2021. Studies in Computational Intelligence Vol. 1013 (eds Shaban-Nejad, A. et al.) 281–293 (Springer, 2022). source
• Not checked yetcites [197] Kwak, H. et al. Drug-disease graph: predicting adverse drug reaction signals via graph neural network with clinical data. In Pacific-Asia Conference on Knowledge Discovery and Data Mining 633–644 (Springer-Verlag, 2020). source

In the paper

“Rather than assuming a certain structure in the EHRs, a Graph Convolution Transformer can even learn the hidden EHR structure [79].”

What the sources say

• Not checked yetcites [79] Schütt, K. et al. Schnet: a continuous-filter convolutional neural network for modeling quantum interactions. In 31st Conference on Neural Information Processing Systems (eds Guyon, I. et al.) (Curran Associates, 2017).

In the paper

“Electronic health records are typically represented by ICD (International Classification of Disease) codes [186, 187].”

What the sources say

• Not checked yetcites [186] Hu, J. et al. SpaGCN: integrating gene expression, spatial location and histology to identify spatial domains and spatially variable genes by graph convolutional network. Nat. Methods 18, 1342–1351 (2021). source
• Not checked yetcites [187] Rotmensch, M., Halpern, Y., Tlimat, A., Horng, S. & Sontag, D. Learning a health knowledge graph from electronic medical records. Sci. Rep. 7, 5994 (2017). source

In the paper

“A mixed pooling multi-view self-attention autoencoder has generated patient representations for predicting either a patient’s risk of developing a disease in a future visit, or the diagnostic codes of the next visit [207].”

What the sources say

• Not checked yetcites [207] Hovenga, V., Oluwadare, O. & Kalita, J. Hic-GNN: a generalizable model for 3D chromosome reconstruction using graph convolutional neural networks. Preprint at bioRxiv https://doi.org/10.1101/2021.11.29.470405 (2021). source

In the paper

“By re-imagining GWAS and expression Quantitative Trait Loci (eQTL) studies [218] as networks, we can already begin to discover biologically meaningful modules to highlight key genes involved in the underlying mechanisms of a disease [219].”

What the sources say

• Not checked yetcites [219] Li, M. M. & Zitnik, M. Deep contextual learners for protein networks. In International Conference on Machine Learning Workshop on Computational Biology (2021).

In the paper

“A combined LSTM and GNN model has also been used to represent patient status sequences and temporal medical event graphs, respectively, to predict future prescriptions or disease codes (Section 2.2) [208, 209].”

What the sources say

• Not checked yetcites [208] Ding, J., Sharon, N. & Bar-Joseph, Z. Temporal modelling using single-cell transcriptomics. Nat. Rev. Genet. 23, 355–368 (2022). source
• Not checked yetcites [209] Fortelny, N. & Bock, C. Knowledge-primed neural networks enable biologically interpretable deep learning on single-cell sequencing data. Genome Biol 21, 190 (2020). source

In the paper

“Effective integration of healthcare data with knowledge about molecular, genomic, disease-level, and drug-level data can help generate more accurate and interpretable predictions about the biological systems of health and disease [224].”

What the sources say

• Not checked yetcites [224] Alessandri, L. et al. Sparsely-connected autoencoder (SCA) for single cell RNAseq data mining. npj Syst. Biol. Appl. 7, 1 (2021). source

In the paper

“With spatial gene expression graphs (weighted and undirected) and corresponding histopathology images, gene expression information are aggregated to generate embeddings of genes that could then be used to investigate spatial domains (e.g., differentiate between cancer and noncancer regions in tissues) [200].”

What the sources say

• Not checked yetcites [200] Hosseini, A., Chen, T., Wu, W., Sun, Y. & Sarrafzadeh, M. HeteroMed: heterogeneous information network for medical diagnosis. In International Conference on Information and Knowledge Management 763–772 (ACM, 2018). source

In the paper

“GNNs are able to model relationships between lymph nodes to compute the spread of lymph node gross tumor volume based on radiotherapy CT images (Section 2.2) [194].”

What the sources say

• Not checked yetcites [194] Liu, S. et al. A hybrid method of recurrent neural network and graph neural network for next-period prescription prediction. Int. J. Mach. Learn. Cybern. 11, 2849–2856 (2020). source

In the paper

“Low dimensional embeddings of EHR data can be generated by separately considering medical services, doctors, and patients in shallow network embeddings (Section 2.1) and graph neural networks (Section 2.2) [202, 203].”

What the sources say

• Not checked yetcites [202] Wu, S., Chen, D. & Snyder, M. P. Network biology bridges the gaps between quantitative genetics and multi-omics to map complex diseases. Curr. Opin. Chem. Biol. 66, 102101 (2022). source
• Not checked yetcites [203] Umans, B. D., Battle, A. & Gilad, Y. Where are the disease-associated eQTLs? Trends Genet 37, 109–124 (2020). source

In the paper

“GNNs are also shown to leverage relational structures like similarities among chest X-rays to improve downstream tasks, such as disease diagnosis and localization [198].”

What the sources say

• Not checked yetcites [198] Zhao, C., Jiang, J., Guan, Y., Guo, X. & He, B. EMR-based medical knowledge representation and inference via Markov random fields and distributed representation learning. Artif. Intell. Med. 87, 49–59 (2018). source

In the paper

“By integrating patient records into our networks, graph representation learning is well-equipped to advance precision medicine by generating predictions tailored to individual patients.”

In the paper

“Moreover, combined RNNs and GNNs models have been applied to EHR data integrated with drug and disease interactions to better recommend medication combinations [215].”

What the sources say

• Not checked yetcites [215] Xu, K. et al. Representation learning on graphs with jumping knowledge networks. In Proc. 35th International Conference on MachineLearning (PMLR, 2018).

In the paper

“Methods that embed medical entities, including EHRs and medical ontologies, leverage the inherently hierarchical structure in the medical concepts KG [201].”

What the sources say

• Not checked yetcites [201] Shang, J., Xiao, C., Ma, T., Li, H. & Sun, J. GameNet: graph augmented memory networks for recommending medication combination. In Proc. AAAI Conference on Artificial Intelligence 1126–1133 (AAAI, 2019). source