Deriving disease modules from the compressed transcriptional space embedded in a deep autoencoder

CGMega: explainable graph neural network framework with attention mechanisms for cancer gene module dissection

Cancer is rarely the straightforward consequence of an abnormality in a single gene, but rather reflects a complex interplay of many genes, represented as gene modules. Here, we leverage the recent advances of model-agnostic interpretation approach and develop CGMega, an explainable and graph attention-based deep learning framework to perform cancer gene module dissection. CGMega outperforms current approaches in cancer gene prediction, and it provides a promising approach to integrate multi-omics information. We apply CGMega to breast cancer cell line and acute myeloid leukemia (AML) patients, and we uncover the high-order gene module formed by ErbB family and tumor factors NRG1, PPM1A and DLG2. We identify 396 candidate AML genes, and observe the enrichment of either known AML genes or candidate AML genes in a single gene module. We also identify patient-specific AML genes and associated gene modules. Together, these results indicate that CGMega can be used to dissect cancer gene modules, and provide high-order mechanistic insights into cancer development and heterogeneity.

Structure-primed embedding on the transcription factor manifold enables transparent model architectures for gene regulatory network and latent activity inference

BACKGROUND

Modeling of gene regulatory networks (GRNs) is limited due to a lack of direct measurements of genome-wide transcription factor activity (TFA) making it difficult to separate covariance and regulatory interactions. Inference of regulatory interactions and TFA requires aggregation of complementary evidence. Estimating TFA explicitly is problematic as it disconnects GRN inference and TFA estimation and is unable to account for, for example, contextual transcription factor-transcription factor interactions, and other higher order features. Deep-learning offers a potential solution, as it can model complex interactions and higher-order latent features, although does not provide interpretable models and latent features.

RESULTS

We propose a novel autoencoder-based framework, StrUcture Primed Inference of Regulation using latent Factor ACTivity (SupirFactor) for modeling, and a metric, explained relative variance (ERV), for interpretation of GRNs. We evaluate SupirFactor with ERV in a wide set of contexts. Compared to current state-of-the-art GRN inference methods, SupirFactor performs favorably. We evaluate latent feature activity as an estimate of TFA and biological function in S. cerevisiae as well as in peripheral blood mononuclear cells (PBMC).

CONCLUSION

Here we present a framework for structure-primed inference and interpretation of GRNs, SupirFactor, demonstrating interpretability using ERV in multiple biological and experimental settings. SupirFactor enables TFA estimation and pathway analysis using latent factor activity, demonstrated here on two large-scale single-cell datasets, modeling S. cerevisiae and PBMC. We find that the SupirFactor model facilitates biological analysis acquiring novel functional and regulatory insight.

XA4C: eXplainable representation learning via Autoencoders revealing Critical genes

Machine Learning models have been frequently used in transcriptome analyses. Particularly, Representation Learning (RL), e.g., autoencoders, are effective in learning critical representations in noisy data. However, learned representations, e.g., the "latent variables" in an autoencoder, are difficult to interpret, not to mention prioritizing essential genes for functional follow-up. In contrast, in traditional analyses, one may identify important genes such as Differentially Expressed (DiffEx), Differentially Co-Expressed (DiffCoEx), and Hub genes. Intuitively, the complex gene-gene interactions may be beyond the capture of marginal effects (DiffEx) or correlations (DiffCoEx and Hub), indicating the need of powerful RL models. However, the lack of interpretability and individual target genes is an obstacle for RL's broad use in practice. To facilitate interpretable analysis and gene-identification using RL, we propose "Critical genes", defined as genes that contribute highly to learned representations (e.g., latent variables in an autoencoder). As a proof-of-concept, supported by eXplainable Artificial Intelligence (XAI), we implemented eXplainable Autoencoder for Critical genes (XA4C) that quantifies each gene's contribution to latent variables, based on which Critical genes are prioritized. Applying XA4C to gene expression data in six cancers showed that Critical genes capture essential pathways underlying cancers. Remarkably, Critical genes has little overlap with Hub or DiffEx genes, however, has a higher enrichment in a comprehensive disease gene database (DisGeNET) and a cancer-specific database (COSMIC), evidencing its potential to disclose massive unknown biology. As an example, we discovered five Critical genes sitting in the center of Lysine degradation (hsa00310) pathway, displaying distinct interaction patterns in tumor and normal tissues. In conclusion, XA4C facilitates explainable analysis using RL and Critical genes discovered by explainable RL empowers the study of complex interactions.

NCAE: data-driven representations using a deep network-coherent DNA methylation autoencoder identify robust disease and risk factor signatures

Precision medicine relies on the identification of robust disease and risk factor signatures from omics data. However, current knowledge-driven approaches may overlook novel or unexpected phenomena due to the inherent biases in biological knowledge. In this study, we present a data-driven signature discovery workflow for DNA methylation analysis utilizing network-coherent autoencoders (NCAEs) with biologically relevant latent embeddings. First, we explored the architecture space of autoencoders trained on a large-scale pan-tissue compendium (n = 75 272) of human epigenome-wide association studies. We observed the emergence of co-localized patterns in the deep autoencoder latent space representations that corresponded to biological network modules. We determined the NCAE configuration with the strongest co-localization and centrality signals in the human protein interactome. Leveraging the NCAE embeddings, we then trained interpretable deep neural networks for risk factor (aging, smoking) and disease (systemic lupus erythematosus) prediction and classification tasks. Remarkably, our NCAE embedding-based models outperformed existing predictors, revealing novel DNA methylation signatures enriched in gene sets and pathways associated with the studied condition in each case. Our data-driven biomarker discovery workflow provides a generally applicable pipeline to capture relevant risk factor and disease information. By surpassing the limitations of knowledge-driven methods, our approach enhances the understanding of complex epigenetic processes, facilitating the development of more effective diagnostic and therapeutic strategies.

An Overview of Deep Generative Models in Functional and Evolutionary Genomics

Following the widespread use of deep learning for genomics, deep generative modeling is also becoming a viable methodology for the broad field. Deep generative models (DGMs) can learn the complex structure of genomic data and allow researchers to generate novel genomic instances that retain the real characteristics of the original dataset. Aside from data generation, DGMs can also be used for dimensionality reduction by mapping the data space to a latent space, as well as for prediction tasks via exploitation of this learned mapping or supervised/semi-supervised DGM designs. In this review, we briefly introduce generative modeling and two currently prevailing architectures, we present conceptual applications along with notable examples in functional and evolutionary genomics, and we provide our perspective on potential challenges and future directions.

Predicting molecular mechanisms of hereditary diseases by using their tissue-selective manifestation

How do aberrations in widely expressed genes lead to tissue-selective hereditary diseases? Previous attempts to answer this question were limited to testing a few candidate mechanisms. To answer this question at a larger scale, we developed "Tissue Risk Assessment of Causality by Expression" (TRACE), a machine learning approach to predict genes that underlie tissue-selective diseases and selectivity-related features. TRACE utilized 4,744 biologically interpretable tissue-specific gene features that were inferred from heterogeneous omics datasets. Application of TRACE to 1,031 disease genes uncovered known and novel selectivity-related features, the most common of which was previously overlooked. Next, we created a catalog of tissue-associated risks for 18,927 protein-coding genes (https://netbio.bgu.ac.il/trace/). As proof-of-concept, we prioritized candidate disease genes identified in 48 rare-disease patients. TRACE ranked the verified disease gene among the patient's candidate genes significantly better than gene prioritization methods that rank by gene constraint or tissue expression. Thus, tissue selectivity combined with machine learning enhances genetic and clinical understanding of hereditary diseases.

Machine learning in rare disease

High-throughput profiling methods (such as genomics or imaging) have accelerated basic research and made deep molecular characterization of patient samples routine. These approaches provide a rich portrait of genes, molecular pathways and cell types involved in disease phenotypes. Machine learning (ML) can be a useful tool for extracting disease-relevant patterns from high-dimensional datasets. However, depending upon the complexity of the biological question, machine learning often requires many samples to identify recurrent and biologically meaningful patterns. Rare diseases are inherently limited in clinical cases, leading to few samples to study. In this Perspective, we outline the challenges and emerging solutions for using ML for small sample sets, specifically in rare diseases. Advances in ML methods for rare diseases are likely to be informative for applications beyond rare diseases for which few samples exist with high-dimensional data. We propose that the method community prioritize the development of ML techniques for rare disease research.

siVAE: interpretable deep generative models for single-cell transcriptomes

Neural networks such as variational autoencoders (VAE) perform dimensionality reduction for the visualization and analysis of genomic data, but are limited in their interpretability: it is unknown which data features are represented by each embedding dimension. We present siVAE, a VAE that is interpretable by design, thereby enhancing downstream analysis tasks. Through interpretation, siVAE also identifies gene modules and hubs without explicit gene network inference. We use siVAE to identify gene modules whose connectivity is associated with diverse phenotypes such as iPSC neuronal differentiation efficiency and dementia, showcasing the wide applicability of interpretable generative models for genomic data analysis.

Structure primed embedding on the transcription factor manifold enables transparent model architectures for gene regulatory network and latent activity inference

The modeling of gene regulatory networks (GRNs) is limited due to a lack of direct measurements of regulatory features in genome-wide screens. Most GRN inference methods are therefore forced to model relationships between regulatory genes and their targets with expression as a proxy for the upstream independent features, complicating validation and predictions produced by modeling frameworks. Separating covariance and regulatory influence requires aggregation of independent and complementary sets of evidence, such as transcription factor (TF) binding and target gene expression. However, the complete regulatory state of the system, e.g. TF activity (TFA) is unknown due to a lack of experimental feasibility, making regulatory relations difficult to infer. Some methods attempt to account for this by modeling TFA as a latent feature, but these models often use linear frameworks that are unable to account for non-linearities such as saturation, TF-TF interactions, and other higher order features. Deep learning frameworks may offer a solution, as they are capable of modeling complex interactions and capturing higher-order latent features. However, these methods often discard central concepts in biological systems modeling, such as sparsity and latent feature interpretability, in favor of increased model complexity. We propose a novel deep learning autoencoder-based framework, StrUcture Primed Inference of Regulation using latent Factor ACTivity (SupirFactor), that scales to single cell genomic data and maintains interpretability to perform GRN inference and estimate TFA as a latent feature. We demonstrate that SupirFactor outperforms current leading GRN inference methods, predicts biologically relevant TFA and elucidates functional regulatory pathways through aggregation of TFs.

Animal models and animal-free innovations for cardiovascular research: current status and routes to be explored. Consensus document of the ESC Working Group on Myocardial Function and the ESC Working Group on Cellular Biology of the Heart

Cardiovascular diseases represent a major cause of morbidity and mortality, necessitating research to improve diagnostics, and to discover and test novel preventive and curative therapies, all of which warrant experimental models that recapitulate human disease. The translation of basic science results to clinical practice is a challenging task, in particular for complex conditions such as cardiovascular diseases, which often result from multiple risk factors and comorbidities. This difficulty might lead some individuals to question the value of animal research, citing the translational 'valley of death', which largely reflects the fact that studies in rodents are difficult to translate to humans. This is also influenced by the fact that new, human-derived in vitro models can recapitulate aspects of disease processes. However, it would be a mistake to think that animal models do not represent a vital step in the translational pathway as they do provide important pathophysiological insights into disease mechanisms particularly on an organ and systemic level. While stem cell-derived human models have the potential to become key in testing toxicity and effectiveness of new drugs, we need to be realistic, and carefully validate all new human-like disease models. In this position paper, we highlight recent advances in trying to reduce the number of animals for cardiovascular research ranging from stem cell-derived models to in situ modelling of heart properties, bioinformatic models based on large datasets, and state-of-the-art animal models, which show clinically relevant characteristics observed in patients with a cardiovascular disease. We aim to provide a guide to help researchers in their experimental design to translate bench findings to clinical routine taking the replacement, reduction, and refinement (3R) as a guiding concept.

SD2: spatially resolved transcriptomics deconvolution through integration of dropout and spatial information

MOTIVATION

Unveiling the heterogeneity in the tissues is crucial to explore cell-cell interactions and cellular targets of human diseases. Spatial transcriptomics (ST) supplies spatial gene expression profile which has revolutionized our biological understanding, but variations in cell-type proportions of each spot with dozens of cells would confound downstream analysis. Therefore, deconvolution of ST has been an indispensable step and a technical challenge toward the higher-resolution panorama of tissues.

RESULTS

Here, we propose a novel ST deconvolution method called SD2 integrating spatial information of ST data and embracing an important characteristic, dropout, which is traditionally considered as an obstruction in single-cell RNA sequencing data (scRNA-seq) analysis. First, we extract the dropout-based genes as informative features from ST and scRNA-seq data by fitting a Michaelis-Menten function. After synthesizing pseudo-ST spots by randomly composing cells from scRNA-seq data, auto-encoder is applied to discover low-dimensional and non-linear representation of the real- and pseudo-ST spots. Next, we create a graph containing embedded profiles as nodes, and edges determined by transcriptional similarity and spatial relationship. Given the graph, a graph convolutional neural network is used to predict the cell-type compositions for real-ST spots. We benchmark the performance of SD2 on the simulated seqFISH+ dataset with different resolutions and measurements which show superior performance compared with the state-of-the-art methods. SD2 is further validated on three real-world datasets with different ST technologies and demonstrates the capability to localize cell-type composition accurately with quantitative evidence. Finally, ablation study is conducted to verify the contribution of different modules proposed in SD2.

AVAILABILITY AND IMPLEMENTATION

The SD2 is freely available in github (https://github.com/leihouyeung/SD2) and Zenodo (https://doi.org/10.5281/zenodo.7024684).

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

MOMA: a multi-task attention learning algorithm for multi-omics data interpretation and classification

MOTIVATION

Accurate diagnostic classification and biological interpretation are important in biology and medicine, which are data-rich sciences. Thus, integration of different data types is necessary for the high predictive accuracy of clinical phenotypes, and more comprehensive analyses for predicting the prognosis of complex diseases are required.

RESULTS

Here, we propose a novel multi-task attention learning algorithm for multi-omics data, termed MOMA, which captures important biological processes for high diagnostic performance and interpretability. MOMA vectorizes features and modules using a geometric approach and focuses on important modules in multi-omics data via an attention mechanism. Experiments using public data on Alzheimer's disease and cancer with various classification tasks demonstrated the superior performance of this approach. The utility of MOMA was also verified using a comparison experiment with an attention mechanism that was turned on or off and biological analysis.

AVAILABILITY AND IMPLEMENTATION

The source codes are available at https://github.com/dmcb-gist/MOMA.

SUPPLEMENTARY INFORMATION

Supplementary materials are available at Bioinformatics online.

HyMM: hybrid method for disease-gene prediction by integrating multiscale module structure

MOTIVATION

Identifying disease-related genes is an important issue in computational biology. Module structure widely exists in biomolecule networks, and complex diseases are usually thought to be caused by perturbations of local neighborhoods in the networks, which can provide useful insights for the study of disease-related genes. However, the mining and effective utilization of the module structure is still challenging in such issues as a disease gene prediction.

RESULTS

We propose a hybrid disease-gene prediction method integrating multiscale module structure (HyMM), which can utilize multiscale information from local to global structure to more effectively predict disease-related genes. HyMM extracts module partitions from local to global scales by multiscale modularity optimization with exponential sampling, and estimates the disease relatedness of genes in partitions by the abundance of disease-related genes within modules. Then, a probabilistic model for integration of gene rankings is designed in order to integrate multiple predictions derived from multiscale module partitions and network propagation, and a parameter estimation strategy based on functional information is proposed to further enhance HyMM's predictive power. By a series of experiments, we reveal the importance of module partitions at different scales, and verify the stable and good performance of HyMM compared with eight other state-of-the-arts and its further performance improvement derived from the parameter estimation.

CONCLUSIONS

The results confirm that HyMM is an effective framework for integrating multiscale module structure to enhance the ability to predict disease-related genes, which may provide useful insights for the study of the multiscale module structure and its application in such issues as a disease-gene prediction.

Deep neural network prediction of genome-wide transcriptome signatures - beyond the Black-box

Prediction algorithms for protein or gene structures, including transcription factor binding from sequence information, have been transformative in understanding gene regulation. Here we ask whether human transcriptomic profiles can be predicted solely from the expression of transcription factors (TFs). We find that the expression of 1600 TFs can explain >95% of the variance in 25,000 genes. Using the light-up technique to inspect the trained NN, we find an over-representation of known TF-gene regulations. Furthermore, the learned prediction network has a hierarchical organization. A smaller set of around 125 core TFs could explain close to 80% of the variance. Interestingly, reducing the number of TFs below 500 induces a rapid decline in prediction performance. Next, we evaluated the prediction model using transcriptional data from 22 human diseases. The TFs were sufficient to predict the dysregulation of the target genes (rho = 0.61, P < 10⁻²¹⁶). By inspecting the model, key causative TFs could be extracted for subsequent validation using disease-associated genetic variants. We demonstrate a methodology for constructing an interpretable neural network predictor, where analyses of the predictors identified key TFs that were inducing transcriptional changes during disease.

Identifying Breast Cancer-Related Genes Based on a Novel Computational Framework Involving KEGG Pathways and PPI Network Modularity

Complex diseases, such as breast cancer, are often caused by mutations of multiple functional genes. Identifying disease-related genes is a critical and challenging task for unveiling the biological mechanisms behind these diseases. In this study, we develop a novel computational framework to analyze the network properties of the known breast cancer–associated genes, based on which we develop a random-walk-with-restart (RCRWR) algorithm to predict novel disease genes. Specifically, we first curated a set of breast cancer–associated genes from the Genome-Wide Association Studies catalog and Online Mendelian Inheritance in Man database and then studied the distribution of these genes on an integrated protein–protein interaction (PPI) network. We found that the breast cancer–associated genes are significantly closer to each other than random, which confirms the modularity property of disease genes in a PPI network as revealed by previous studies. We then retrieved PPI subnetworks spanning top breast cancer–associated KEGG pathways and found that the distribution of these genes on the subnetworks are non-random, suggesting that these KEGG pathways are activated non-uniformly. Taking advantage of the non-random distribution of breast cancer–associated genes, we developed an improved RCRWR algorithm to predict novel cancer genes, which integrates network reconstruction based on local random walk dynamics and subnetworks spanning KEGG pathways. Compared with the disease gene prediction without using the information from the KEGG pathways, this method has a better prediction performance on inferring breast cancer–associated genes, and the top predicted genes are better enriched on known breast cancer–associated gene ontologies. Finally, we performed a literature search on top predicted novel genes and found that most of them are supported by at least wet-lab experiments on cell lines. In summary, we propose a robust computational framework to prioritize novel breast cancer–associated genes, which could be used for further in vitro and in vivo experimental validation.

Identifying Breast Cancer-Related Genes Based on a Novel Computational Framework Involving KEGG Pathways and PPI Network Modularity

Complex diseases, such as breast cancer, are often caused by mutations of multiple functional genes. Identifying disease-related genes is a critical and challenging task for unveiling the biological mechanisms behind these diseases. In this study, we develop a novel computational framework to analyze the network properties of the known breast cancer-associated genes, based on which we develop a random-walk-with-restart (RCRWR) algorithm to predict novel disease genes. Specifically, we first curated a set of breast cancer-associated genes from the Genome-Wide Association Studies catalog and Online Mendelian Inheritance in Man database and then studied the distribution of these genes on an integrated protein-protein interaction (PPI) network. We found that the breast cancer-associated genes are significantly closer to each other than random, which confirms the modularity property of disease genes in a PPI network as revealed by previous studies. We then retrieved PPI subnetworks spanning top breast cancer-associated KEGG pathways and found that the distribution of these genes on the subnetworks are non-random, suggesting that these KEGG pathways are activated non-uniformly. Taking advantage of the non-random distribution of breast cancer-associated genes, we developed an improved RCRWR algorithm to predict novel cancer genes, which integrates network reconstruction based on local random walk dynamics and subnetworks spanning KEGG pathways. Compared with the disease gene prediction without using the information from the KEGG pathways, this method has a better prediction performance on inferring breast cancer-associated genes, and the top predicted genes are better enriched on known breast cancer-associated gene ontologies. Finally, we performed a literature search on top predicted novel genes and found that most of them are supported by at least wet-lab experiments on cell lines. In summary, we propose a robust computational framework to prioritize novel breast cancer-associated genes, which could be used for further in vitro and in vivo experimental validation.

A deep learning method for HLA imputation and trans-ethnic MHC fine-mapping of type 1 diabetes

Conventional human leukocyte antigen (HLA) imputation methods drop their performance for infrequent alleles, which is one of the factors that reduce the reliability of trans-ethnic major histocompatibility complex (MHC) fine-mapping due to inter-ethnic heterogeneity in allele frequency spectra. We develop DEEP*HLA, a deep learning method for imputing HLA genotypes. Through validation using the Japanese and European HLA reference panels (n = 1,118 and 5,122), DEEP*HLA achieves the highest accuracies with significant superiority for low-frequency and rare alleles. DEEP*HLA is less dependent on distance-dependent linkage disequilibrium decay of the target alleles and might capture the complicated region-wide information. We apply DEEP*HLA to type 1 diabetes GWAS data from BioBank Japan (n = 62,387) and UK Biobank (n = 354,459), and successfully disentangle independently associated class I and II HLA variants with shared risk among diverse populations (the top signal at amino acid position 71 of HLA-DRβ1; P = 7.5 × 10^-120). Our study illustrates the value of deep learning in genotype imputation and trans-ethnic MHC fine-mapping.

Domain randomization-enhanced deep learning models for bird detection

Automatic bird detection in ornithological analyses is limited by the accuracy of existing models, due to the lack of training data and the difficulties in extracting the fine-grained features required to distinguish bird species. Here we apply the domain randomization strategy to enhance the accuracy of the deep learning models in bird detection. Trained with virtual birds of sufficient variations in different environments, the model tends to focus on the fine-grained features of birds and achieves higher accuracies. Based on the 100 terabytes of 2-month continuous monitoring data of egrets, our results cover the findings using conventional manual observations, e.g., vertical stratification of egrets according to body size, and also open up opportunities of long-term bird surveys requiring intensive monitoring that is impractical using conventional methods, e.g., the weather influences on egrets, and the relationship of the migration schedules between the great egrets and little egrets.

Biologically Informed Neural Networks Predict Drug Responses

Deep neural networks often achieve high predictive accuracy on biological problems, but it can be hard to contextualize how and explain why predictions are made. In this issue, Kuenzi et al. model the sensitivity of cancers to drugs using deep neural networks with a hierarchical structure derived from the Gene Ontology.