GLDM: hit molecule generation with constrained graph latent diffusion model.

Discovering hit molecules with desired biological activity in a directed manner is a promising but profound task in computer-aided drug discovery. Inspired by recent generative AI approaches, particularly Diffusion Models (DM), we propose Graph Latent Diffusion Model (GLDM)-a latent DM that preserves both the effectiveness of autoencoders of compressing complex chemical data and the DM's capabilities of generating novel molecules. Specifically, we first develop an autoencoder to encode the molecular data into low-dimensional latent representations and then train the DM on the latent space to generate molecules inducing targeted biological activity defined by gene expression profiles. Manipulating DM in the latent space rather than the input space avoids complicated operations to map molecule decomposition and reconstruction to diffusion processes, and thus improves training efficiency. Experiments show that GLDM not only achieves outstanding performances on molecular generation benchmarks, but also generates samples with optimal chemical properties and potentials to induce desired biological activity.

DrDimont: explainable drug response prediction from differential analysis of multi-omics networks.

MOTIVATION: While it has been well established that drugs affect and help patients differently, personalized drug response predictions remain challenging. Solutions based on single omics measurements have been proposed, and networks provide means to incorporate molecular interactions into reasoning. However, how to integrate the wealth of information contained in multiple omics layers still poses a complex problem. RESULTS: We present DrDimont, Drug response prediction from Differential analysis of multi-omics networks. It allows for comparative conclusions between two conditions and translates them into differential drug response predictions. DrDimont focuses on molecular interactions. It establishes condition-specific networks from correlation within an omics layer that are then reduced and combined into heterogeneous, multi-omics molecular networks. A novel semi-local, path-based integration step ensures integrative conclusions. Differential predictions are derived from comparing the condition-specific integrated networks. DrDimont's predictions are explainable, i.e. molecular differences that are the source of high differential drug scores can be retrieved. We predict differential drug response in breast cancer using transcriptomics, proteomics, phosphosite and metabolomics measurements and contrast estrogen receptor positive and receptor negative patients. DrDimont performs better than drug prediction based on differential protein expression or PageRank when evaluating it on ground truth data from cancer cell lines. We find proteomic and phosphosite layers to carry most information for distinguishing drug response. AVAILABILITY AND IMPLEMENTATION: DrDimont is available on CRAN: https://cran.r-project.org/package=DrDimont. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Deep learning and multi-omics approach to predict drug responses in cancer.

BACKGROUND: Cancers are genetically heterogeneous, so anticancer drugs show varying degrees of effectiveness on patients due to their differing genetic profiles. Knowing patient's responses to numerous cancer drugs are needed for personalized treatment for cancer. By using molecular profiles of cancer cell lines available from Cancer Cell Line Encyclopedia (CCLE) and anticancer drug responses available in the Genomics of Drug Sensitivity in Cancer (GDSC), we will build computational models to predict anticancer drug responses from molecular features. RESULTS: We propose a novel deep neural network model that integrates multi-omics data available as gene expressions, copy number variations, gene mutations, reverse phase protein array expressions, and metabolomics expressions, in order to predict cellular responses to known anti-cancer drugs. We employ a novel graph embedding layer that incorporates interactome data as prior information for prediction. Moreover, we propose a novel attention layer that effectively combines different omics features, taking their interactions into account. The network outperformed feedforward neural networks and reported 0.90 for [Formula: see text] values for prediction of drug responses from cancer cell lines data available in CCLE and GDSC. CONCLUSION: The outstanding results of our experiments demonstrate that the proposed method is capable of capturing the interactions of genes and proteins, and integrating multi-omics features effectively. Furthermore, both the results of ablation studies and the investigations of the attention layer imply that gene mutation has a greater influence on the prediction of drug responses than other omics data types. Therefore, we conclude that our approach can not only predict the anti-cancer drug response precisely but also provides insights into reaction mechanisms of cancer cell lines and drugs as well.

Decoding task specific and task general functional architectures of the brain.

Functional magnetic resonance imaging (fMRI) is used to capture complex and dynamic interactions between brain regions while performing tasks. Task related alterations in the brain have been classified as task specific and task general, depending on whether they are particular to a task or common across multiple tasks. Using recent attempts in interpreting deep learning models, we propose an approach to determine both task specific and task general architectures of the functional brain. We demonstrate our methods with a reference-based decoder on deep learning classifiers trained on 12,500 rest and task fMRI samples from the Human Connectome Project (HCP). The decoded task general and task specific motor and language architectures were validated with findings from previous studies. We found that unlike intersubject variability that is characteristic of functional pathology of neurological diseases, a small set of connections are sufficient to delineate the rest and task states. The nodes and connections in the task general architecture could serve as potential disease biomarkers as alterations in task general brain modulations are known to be implicated in several neuropsychiatric disorders.

Graph embeddings on gene ontology annotations for protein-protein interaction prediction.

BACKGROUND: Protein-protein interaction (PPI) prediction is an important task towards the understanding of many bioinformatics functions and applications, such as predicting protein functions, gene-disease associations and disease-drug associations. However, many previous PPI prediction researches do not consider missing and spurious interactions inherent in PPI networks. To address these two issues, we define two corresponding tasks, namely missing PPI prediction and spurious PPI prediction, and propose a method that employs graph embeddings that learn vector representations from constructed Gene Ontology Annotation (GOA) graphs and then use embedded vectors to achieve the two tasks. Our method leverages on information from both term-term relations among GO terms and term-protein annotations between GO terms and proteins, and preserves properties of both local and global structural information of the GO annotation graph. RESULTS: We compare our method with those methods that are based on information content (IC) and one method that is based on word embeddings, with experiments on three PPI datasets from STRING database. Experimental results demonstrate that our method is more effective than those compared methods. CONCLUSION: Our experimental results demonstrate the effectiveness of using graph embeddings to learn vector representations from undirected GOA graphs for our defined missing and spurious PPI tasks.

Functional homogeneity and specificity of topological modules in human proteome.

BACKGROUND: Functional modules in protein-protein interaction networks (PPIN) are defined by maximal sets of functionally associated proteins and are vital to understanding cellular mechanisms and identifying disease associated proteins. Topological modules of the human proteome have been shown to be related to functional modules of PPIN. However, the effects of the weights of interactions between protein pairs and the integration of physical (direct) interactions with functional (indirect expression-based) interactions have not been investigated in the detection of functional modules of the human proteome. RESULTS: We investigated functional homogeneity and specificity of topological modules of the human proteome and validated them with known biological and disease pathways. Specifically, we determined the effects on functional homogeneity and heterogeneity of topological modules (i) with both physical and functional protein-protein interactions; and (ii) with incorporation of functional similarities between proteins as weights of interactions. With functional enrichment analyses and a novel measure for functional specificity, we evaluated functional relevance and specificity of topological modules of the human proteome. CONCLUSIONS: The topological modules ranked using specificity scores show high enrichment with gene sets of known functions. Physical interactions in PPIN contribute to high specificity of the topological modules of the human proteome whereas functional interactions contribute to high homogeneity of the modules. Weighted networks result in more number of topological modules but did not affect their functional propensity. Modules of human proteome are more homogeneous for molecular functions than biological processes.

Refining modules to determine functionally significant clusters in molecular networks.

BACKGROUND: Module detection algorithms relying on modularity maximization suffer from an inherent resolution limit that hinders detection of small topological modules, especially in molecular networks where most biological processes are believed to form small and compact communities. We propose a novel modular refinement approach that helps finding functionally significant modules of molecular networks. RESULTS: The module refinement algorithm improves the quality of topological modules in protein-protein interaction networks by finding biologically functionally significant modules. The algorithm is based on the fact that functional modules in biology do not necessarily represent those corresponding to maximum modularity. Larger modules corresponding to maximal modularity are incrementally re-modularized again under specific constraints so that smaller yet topologically and biologically valid modules are recovered. We show improvement in quality and functional coverage of modules using experiments on synthetic and real protein-protein interaction networks. We also compare our results with six existing methods available for clustering biological networks. CONCLUSION: The proposed algorithm finds smaller but functionally relevant modules that are undetected by classical quality maximization approaches for modular detection. The refinement procedure helps to detect more functionally enriched modules in protein-protein interaction networks, which are also more coherent with functionally characterised gene sets.

GO2Vec: transforming GO terms and proteins to vector representations via graph embeddings.

BACKGROUND: Semantic similarity between Gene Ontology (GO) terms is a fundamental measure for many bioinformatics applications, such as determining functional similarity between genes or proteins. Most previous research exploited information content to estimate the semantic similarity between GO terms; recently some research exploited word embeddings to learn vector representations for GO terms from a large-scale corpus. In this paper, we proposed a novel method, named GO2Vec, that exploits graph embeddings to learn vector representations for GO terms from GO graph. GO2Vec combines the information from both GO graph and GO annotations, and its learned vectors can be applied to a variety of bioinformatics applications, such as calculating functional similarity between proteins and predicting protein-protein interactions. RESULTS: We conducted two kinds of experiments to evaluate the quality of GO2Vec: (1) functional similarity between proteins on the Collaborative Evaluation of GO-based Semantic Similarity Measures (CESSM) dataset and (2) prediction of protein-protein interactions on the Yeast and Human datasets from the STRING database. Experimental results demonstrate the effectiveness of GO2Vec over the information content-based measures and the word embedding-based measures. CONCLUSION: Our experimental results demonstrate the effectiveness of using graph embeddings to learn vector representations from undirected GO and GOA graphs. Our results also demonstrate that GO annotations provide useful information for computing the similarity between GO terms and between proteins.

MultiDCoX: Multi-factor analysis of differential co-expression.

BACKGROUND: Differential co-expression (DCX) signifies change in degree of co-expression of a set of genes among different biological conditions. It has been used to identify differential co-expression networks or interactomes. Many algorithms have been developed for single-factor differential co-expression analysis and applied in a variety of studies. However, in many studies, the samples are characterized by multiple factors such as genetic markers, clinical variables and treatments. No algorithm or methodology is available for multi-factor analysis of differential co-expression. RESULTS: We developed a novel formulation and a computationally efficient greedy search algorithm called MultiDCoX to perform multi-factor differential co-expression analysis. Simulated data analysis demonstrates that the algorithm can effectively elicit differentially co-expressed (DCX) gene sets and quantify the influence of each factor on co-expression. MultiDCoX analysis of a breast cancer dataset identified interesting biologically meaningful differentially co-expressed (DCX) gene sets along with genetic and clinical factors that influenced the respective differential co-expression. CONCLUSIONS: MultiDCoX is a space and time efficient procedure to identify differentially co-expressed gene sets and successfully identify influence of individual factors on differential co-expression.

Gene and sample selection using T-score with sample selection.

Gene selection from high-dimensional microarray gene-expression data is statistically a challenging problem. Filter approaches to gene selection have been popular because of their simplicity, efficiency, and accuracy. Due to small sample size, all samples are generally used to compute relevant ranking statistics and selection of samples in filter-based gene selection methods has not been addressed. In this paper, we extend previously-proposed simultaneous sample and gene selection approach. In a backward elimination method, a modified logistic regression loss function is used to select relevant samples at each iteration, and these samples are used to compute the T-score to rank genes. This method provides a compromise solution between T-score and other support vector machine (SVM) based algorithms. The performance is demonstrated on both simulated and real datasets with criteria such as classification performance, stability and redundancy. Results indicate that computational complexity and stability of the method are improved compared to SVM based methods without compromising the classification performance.

Identification of a new export signal in Plasmodium yoelii: identification of a new exportome.

Development of the erythrocytic malaria parasite requires targeting of parasite proteins into multiple compartments located within and beyond the parasite confine. Beyond the PEXEL/VTS pathway and its characterized players, increasing amount of evidence has highlighted the existence of proteins exported using alternative export-signal(s)/pathway(s); hence, the exportomes currently predicted are incomplete. The nature of these exported proteins which could have a prominent role in most of the Plasmodium species remains elusive. Using P. yoelii variant proteins, we identified a signal associated to lipophilic region that mediates export of P. yoelii proteins. This non-PEXEL signal termed PLASMED is defined by semi-conserved residues and possibly a secondary structure. In vivo characterization of exported-proteins indicated that PLASMED is a bona fide export-signal that allowed us to identify an unseen P. yoelii exportome. The repertoire of the newly predicted exported proteins opens up perspectives for unravelling the remodelling of the host-cell by the parasite, against which new therapies could be elaborated.

Extracting rate changes in transcriptional regulation from MEDLINE abstracts.

BACKGROUND: Time delays are important factors that are often neglected in gene regulatory network (GRN) inference models. Validating time delays from knowledge bases is a challenge since the vast majority of biological databases do not record temporal information of gene regulations. Biological knowledge and facts on gene regulations are typically extracted from bio-literature with specialized methods that depend on the regulation task. In this paper, we mine evidences for time delays related to the transcriptional regulation of yeast from the PubMed abstracts. RESULTS: Since the vast majority of abstracts lack quantitative time information, we can only collect qualitative evidences of time delays. Specifically, the speed-up or delay in transcriptional regulation rate can provide evidences for time delays (shorter or longer) in GRN. Thus, we focus on deriving events related to rate changes in transcriptional regulation. A corpus of yeast regulation related abstracts was manually labeled with such events. In order to capture these events automatically, we create an ontology of sub-processes that are likely to result in transcription rate changes by combining textual patterns and biological knowledge. We also propose effective feature extraction methods based on the created ontology to identify the direct evidences with specific details of these events. Our ontologies outperform existing state-of-the-art gene regulation ontologies in the automatic rule learning method applied to our corpus. The proposed deterministic ontology rule-based method can achieve comparable performance to the automatic rule learning method based on decision trees. This demonstrates the effectiveness of our ontology in identifying rate-changing events. We also tested the effectiveness of the proposed feature mining methods on detecting direct evidence of events. Experimental results show that the machine learning method on these features achieves an F1-score of 71.43%. CONCLUSIONS: The manually labeled corpus of events relating to rate changes in transcriptional regulation for yeast is available in https://sites.google.com/site/wentingntu/data. The created ontologies summarized both biological causes of rate changes in transcriptional regulation and corresponding positive and negative textual patterns from the corpus. They are demonstrated to be effective in identifying rate-changing events, which shows the benefits of combining textual patterns and biological knowledge on extracting complex biological events.

qFibrosis: a fully-quantitative innovative method incorporating histological features to facilitate accurate fibrosis scoring in animal model and chronic hepatitis B patients.

BACKGROUND & AIMS: There is increasing need for accurate assessment of liver fibrosis/cirrhosis. We aimed to develop qFibrosis, a fully-automated assessment method combining quantification of histopathological architectural features, to address unmet needs in core biopsy evaluation of fibrosis in chronic hepatitis B (CHB) patients. METHODS: qFibrosis was established as a combined index based on 87 parameters of architectural features. Images acquired from 25 Thioacetamide-treated rat samples and 162 CHB core biopsies were used to train and test qFibrosis and to demonstrate its reproducibility. qFibrosis scoring was analyzed employing Metavir and Ishak fibrosis staging as standard references, and collagen proportionate area (CPA) measurement for comparison. RESULTS: qFibrosis faithfully and reliably recapitulates Metavir fibrosis scores, as it can identify differences between all stages in both animal samples (p<0.001) and human biopsies (p<0.05). It is robust to sampling size, allowing for discrimination of different stages in samples of different sizes (area under the curve (AUC): 0.93-0.99 for animal samples: 1-16 mm(2); AUC: 0.84-0.97 for biopsies: 10-44 mm in length). qFibrosis can significantly predict staging underestimation in suboptimal biopsies (<15 mm) and under- and over-scoring by different pathologists (p<0.001). qFibrosis can also differentiate between Ishak stages 5 and 6 (AUC: 0.73, p=0.008), suggesting the possibility of monitoring intra-stage cirrhosis changes. Best of all, qFibrosis demonstrates superior performance to CPA on all counts. CONCLUSIONS: qFibrosis can improve fibrosis scoring accuracy and throughput, thus allowing for reproducible and reliable analysis of efficacies of anti-fibrotic therapies in clinical research and practice.

Aging exacerbates damage and delays repair of alveolar epithelia following influenza viral pneumonia.

BACKGROUND: Influenza virus infection causes significantly higher levels of morbidity and mortality in the elderly. Studies have shown that impaired immunity in the elderly contributes to the increased susceptibility to influenza virus infection, however, how aging affects the lung tissue damage and repair has not been completely elucidated. METHODS: Aged (16-18 months old) and young (2-3 months old) mice were infected with influenza virus intratracheally. Body weight and mortality were monitored. Different days after infection, lung sections were stained to estimate the overall lung tissue damage and for club cells, pro-SPC+ bronchiolar epithelial cells, alveolar type I and II cells to quantify their frequencies using automated image analysis algorithms. RESULTS: Following influenza infection, aged mice lose more weight and die from otherwise sub-lethal influenza infection in young mice. Although there is no difference in damage and regeneration of club cells between the young and the aged mice, damage to alveolar type I and II cells (AT1s and AT2s) is exacerbated, and regeneration of AT2s and their precursors (pro-SPC-positive bronchiolar epithelial cells) is significantly delayed in the aged mice. We further show that oseltamivir treatment reduces virus load and lung damage, and promotes pulmonary recovery from infection in the aged mice. CONCLUSIONS: These findings show that aging increases susceptibility of the distal lung epithelium to influenza infection and delays the emergence of pro-SPC positive progenitor cells during the repair process. Our findings also shed light on possible approaches to enhance the clinical management of severe influenza pneumonia in the elderly.

Neutrophils infected with highly virulent influenza H3N2 virus exhibit augmented early cell death and rapid induction of type I interferon signaling pathways.

We developed a model of influenza virus infection of neutrophils by inducing differentiation of the MPRO promyelocytic cell line. After 5 days of differentiation, about 20-30% of mature neutrophils could be detected. Only a fraction of neutrophils were infected by highly virulent influenza (HVI) virus, but were unable to support active viral replication compared with MDCK cells. HVI infection of neutrophils augmented early and late apoptosis as indicated by annexin V and TUNEL assays. Comparison between the global transcriptomic responses of neutrophils to HVI and low virulent influenza (LVI) revealed that the IFN regulatory factor and IFN signaling pathways were the most significantly overrepresented pathways, with activation of related genes in HVI as early as 3 h. Relatively consistent results were obtained by real-time RT-PCR of selected genes associated with the type I IFN pathway. Early after HVI infection, comparatively enhanced expression of apoptosis-related genes was also elicited.

Sparse Deep Neural Network for Encoding and Decoding the Structural Connectome.

Brain state classification by applying deep learning techniques on neuroimaging data has become a recent topic of research. However, unlike domains where the data is low dimensional or there are large number of available training samples, neuroimaging data is high dimensional and has few training samples. To tackle these issues, we present a sparse feedforward deep neural architecture for encoding and decoding the structural connectome of the human brain. We use a sparsely connected element-wise multiplication as the first hidden layer and a fixed transform layer as the output layer. The number of trainable parameters and the training time is significantly reduced compared to feedforward networks. We demonstrate superior performance of this architecture in encoding the structural connectome implicated in Alzheimer's disease (AD) and Parkinson's disease (PD) from DTI brain scans. For decoding, we propose recursive feature elimination (RFE) algorithm based on DeepLIFT, layer-wise relevance propagation (LRP), and Integrated Gradients (IG) algorithms to remove irrelevant features and thereby identify key biomarkers associated with AD and PD. We show that the proposed architecture reduces 45.1% and 47.1% of the trainable parameters compared to a feedforward DNN with an increase in accuracy by 2.6 % and 3.1% for cognitively normal (CN) vs AD and CN vs PD classification, respectively. We also show that the proposed RFE method leads to a further increase in accuracy by 2.1% and 4% for CN vs AD and CN vs PD classification, while removing approximately 90% to 95% irrelevant features. Furthermore, we argue that the biomarkers (i.e., key brain regions and connections) identified are consistent with previous literature. We show that relevancy score-based methods can yield high discriminative power and are suitable for brain decoding. We also show that the proposed approach led to a reduction in the number of trainable network parameters, an increase in classification accuracy, and a detection of brain connections and regions that were consistent with earlier studies.

SKGC: A General Semantic-level Knowledge Guided Classification Framework for Fetal Congenital Heart Disease.

Congenital heart disease (CHD) is the most common congenital disability affecting healthy development and growth, even resulting in pregnancy termination or fetal death. Recently, deep learning techniques have made remarkable progress to assist in diagnosing CHD. One very popular method is directly classifying fetal ultrasound images, recognized as abnormal and normal, which tends to focus more on global features and neglects semantic knowledge of anatomical structures. The other approach is segmentation-based diagnosis, which requires a large number of pixel-level annotation masks for training. However, the detailed pixel-level segmentation annotation is costly or even unavailable. Based on the above analysis, we propose SKGC, a universal framework to identify normal or abnormal four-chamber heart (4CH) images, guided by a few annotation masks, while improving accuracy remarkably. SKGC consists of a semantic-level knowledge extraction module (SKEM), a multi-knowledge fusion module (MFM), and a classification module (CM). SKEM is responsible for obtaining high-level semantic knowledge, serving as an abstract representation of the anatomical structures that obstetricians focus on. MFM is a lightweight but efficient module that fuses semantic-level knowledge with the original specific knowledge in ultrasound images. CM classifies the fused knowledge and can be replaced by any advanced classifier. Moreover, we design a new loss function that enhances the constraint between the foreground and background predictions, improving the quality of the semantic-level knowledge. Experimental results on the collected real-world NA-4CH and the publicly FEST datasets show that SKGC achieves impressive performance with the best accuracy of 99.68% and 95.40%, respectively. Notably, the accuracy improves from 74.68% to 88.14% using only 10 labeled masks.

Hybrid UNet transformer architecture for ischemic stoke segmentation with MRI and CT datasets.

A hybrid UNet and Transformer (HUT) network is introduced to combine the merits of the UNet and Transformer architectures, improving brain lesion segmentation from MRI and CT scans. The HUT overcomes the limitations of conventional approaches by utilizing two parallel stages: one based on UNet and the other on Transformers. The Transformer-based stage captures global dependencies and long-range correlations. It uses intermediate feature vectors from the UNet decoder and improves segmentation accuracy by enhancing the attention and relationship modeling between voxel patches derived from the 3D brain volumes. In addition, HUT incorporates self-supervised learning on the transformer network. This allows the transformer network to learn by maintaining consistency between the classification layers of the different resolutions of patches and augmentations. There is an improvement in the rate of convergence of the training and the overall capability of segmentation. Experimental results on benchmark datasets, including ATLAS and ISLES2018, demonstrate HUT's advantage over the state-of-the-art methods. HUT achieves higher Dice scores and reduced Hausdorff Distance scores in single-modality and multi-modality lesion segmentation. HUT outperforms the state-the-art network SPiN in the single-modality MRI segmentation on Anatomical Tracings of lesion After Stroke (ATLAS) dataset by 4.84% of Dice score and a large margin of 40.7% in the Hausdorff Distance score. HUT also performed well on CT perfusion brain scans in the Ischemic Stroke Lesion Segmentation (ISLES2018) dataset and demonstrated an improvement over the recent state-of-the-art network USSLNet by 3.3% in the Dice score and 12.5% in the Hausdorff Distance score. With the analysis of both single and multi-modalities datasets (ATLASR12 and ISLES2018), we show that HUT can perform and generalize well on different datasets. Code is available at: https://github.com/vicsohntu/HUT_CT.

HUT: Hybrid UNet transformer for brain lesion and tumour segmentation.

A supervised deep learning network like the UNet has performed well in segmenting brain anomalies such as lesions and tumours. However, such methods were proposed to perform on single-modality or multi-modality images. We use the Hybrid UNet Transformer (HUT) to improve performance in single-modality lesion segmentation and multi-modality brain tumour segmentation. The HUT consists of two pipelines running in parallel, one of which is UNet-based and the other is Transformer-based. The Transformer-based pipeline relies on feature maps in the intermediate layers of the UNet decoder during training. The HUT network takes in the available modalities of 3D brain volumes and embeds the brain volumes into voxel patches. The transformers in the system improve global attention and long-range correlation between the voxel patches. In addition, we introduce a self-supervised training approach in the HUT framework to enhance the overall segmentation performance. We demonstrate that HUT performs better than the state-of-the-art network SPiN in the single-modality segmentation on Anatomical Tracings of Lesions After Stroke (ATLAS) dataset by 4.84% of Dice score and a significant 41% in the Hausdorff Distance score. HUT also performed well on brain scans in the Brain Tumour Segmentation (BraTS20) dataset and demonstrated an improvement over the state-of-the-art network nnUnet by 0.96% in the Dice score and 4.1% in the Hausdorff Distance score.

Multi-modal graph neural network for early diagnosis of Alzheimer's disease from sMRI and PET scans.

In recent years, deep learning models have been applied to neuroimaging data for early diagnosis of Alzheimer's disease (AD). Structural magnetic resonance imaging (sMRI) and positron emission tomography (PET) images provide structural and functional information about the brain, respectively. Combining these features leads to improved performance than using a single modality alone in building predictive models for AD diagnosis. However, current multi-modal approaches in deep learning, based on sMRI and PET, are mostly limited to convolutional neural networks, which do not facilitate integration of both image and phenotypic information of subjects. We propose to use graph neural networks (GNN) that are designed to deal with problems in non-Euclidean domains. In this study, we demonstrate how brain networks are created from sMRI or PET images and can be used in a population graph framework that combines phenotypic information with imaging features of the brain networks. Then, we present a multi-modal GNN framework where each modality has its own branch of GNN and a technique that combines the multi-modal data at both the level of node vectors and adjacency matrices. Finally, we perform late fusion to combine the preliminary decisions made in each branch and produce a final prediction. As multi-modality data becomes available, multi-source and multi-modal is the trend of AD diagnosis. We conducted explorative experiments based on multi-modal imaging data combined with non-imaging phenotypic information for AD diagnosis and analyzed the impact of phenotypic information on diagnostic performance. Results from experiments demonstrated that our proposed multi-modal approach improves performance for AD diagnosis. Our study also provides technical reference and support the need for multivariate multi-modal diagnosis methods.