With the high prevalence of breast cancer, it is urgent to find out the intrinsic difference between various subtypes, so as to infer the underlying mechanisms. Given the available multi-omics data, their proper integration can improve the accuracy of breast cancer subtype recognition. In this study, DeepMO, a model using deep neural networks based on multi-omics data, was employed for classifying breast cancer subtypes. Three types of omics data including mRNA data, DNA methylation data, and copy number variation (CNV) data were collected from The Cancer Genome Atlas (TCGA). After data preprocessing and feature selection, each type of omics data was input into the deep neural network, which consists of an encoding subnetwork and a classification subnetwork. The results of DeepMO based on multi-omics on binary classification are better than other methods in terms of accuracy and area under the curve (AUC). Moreover, compared with other methods using single omics data and multi-omics data, DeepMO also had a higher prediction accuracy on multi-classification. We also validated the effect of feature selection on DeepMO. Finally, we analyzed the enrichment gene ontology (GO) terms and biological pathways of these significant genes, which were discovered during the feature selection process. We believe that the proposed model is useful for multi-omics data analysis.

Multi-view clustering (MVC), which integrates information from multiple views to enhance performance, has garnered increasing attention in recent years. Partially View-aligned Clustering (PVC), which is a particularly critical aspect of this process, requires a thorough exploration of complementary and consistent information under conditions of partial view alignment. However, most existing PVC methods primarily focus on semantic consistency, employing semantic consistency features for both view alignment and clustering tasks. These methods neglect the effects of noise and complementary information across multiple views and the suitability of these features for clustering. To address these limitations, our approach aims to leverage three distinct types of consistency to extract semantic consistency features and clustering consistency features, which are specifically designed for view alignment and clustering tasks, respectively. By omitting the reconstruction process, we mitigate the adverse effects of mutual information and noise on view alignment. Specifically, we first exploit the structural consistency of similarity graphs across different views to guide feature extraction in view-specific autoencoders. This process produces structural consistency features that are both cluster-discriminative and structurally coherent. Subsequently, two separate multilayer perceptrons (MLPs) are trained via contrastive learning to extract semantic consistency features and clustering consistency features from the structural features. These features are optimized for their respective tasks. Ultimately, a self-paced style view alignment strategy is used to iteratively re-align the data based on semantic and clustering consistency while the model is optimized via the re-aligned data. Extensive experiments on multiple real-world benchmark datasets demonstrate that our method outperforms the state-of-the-art multi-view approaches, highlighting its effectiveness in tackling the challenges of PVC. The code is available at https://github.com/kongyiH/TCLPVC.

We present here an integrated analysis of structures and functions of genome-scale metabolic networks of 17 microorganisms. Our structural analyses of these networks revealed that the node degree of each network, represented as a (simplified) reaction network, follows a power-law distribution, and the clustering coefficient of each network has a positive correlation with the corresponding node degree. Together, these properties imply that each network has exactly one large and densely connected subnetwork or core. Further analyses revealed that each network consists of three functionally distinct subnetworks: (i) a core, consisting of a large number of directed reaction cycles of enzymes for interconversions among intermediate metabolites; (ii) a catabolic module, with a largely layered structure consisting of mostly catabolic enzymes; (iii) an anabolic module with a similar structure consisting of virtually all anabolic genes; and (iv) the three subnetworks cover on average ∼56, ∼31 and ∼13% of a network’s nodes across the 17 networks, respectively. Functional analyses suggest: (1) cellular metabolic fluxes generally go from the catabolic module to the core for substantial interconversions, then the flux directions to anabolic module appear to be determined by input nutrient levels as well as a set of precursors needed for macromolecule syntheses; and (2) enzymes in each subnetwork have characteristic ranges of kinetic parameters, suggesting optimized metabolic and regulatory relationships among the three subnetworks.

CyanoPATH is a database that curates and analyzes the common genomic functional repertoire for cyanobacteria harmful algal blooms (CyanoHABs) in eutrophic waters. Based on the literature of empirical studies and genome/protein databases, it summarizes four types of information: common biological functions (pathways) driving CyanoHABs, customized pathway maps, classification of blooming type based on databases and the genomes of cyanobacteria. A total of 19 pathways are reconstructed, which are involved in the utilization of macronutrients (e.g. carbon, nitrogen, phosphorus and sulfur), micronutrients (e.g. zinc, magnesium, iron, etc.) and other resources (e.g. light and vitamins) and in stress resistance (e.g. lead and copper). These pathways, comprised of both transport and biochemical reactions, are reconstructed with proteins from NCBI and reactions from KEGG and visualized with self-created transport/reaction maps. The pathways are hierarchical and consist of subpathways, protein/enzyme complexes and constituent proteins. New cyanobacterial genomes can be annotated and visualized for these pathways and compared with existing species. This set of genomic functional repertoire is useful in analyzing aquatic metagenomes and metatranscriptomes in CyanoHAB research. Most importantly, it establishes a link between genome and ecology. All these reference proteins, pathways and maps and genomes are free to download at http://www.csbg-jlu.info/CyanoPATH.

Urine excretory proteins are among the most commonly used biomarkers in body fluids. Computational identification of urine excretory proteins can provide very useful information for identifying targeted disease biomarkers in urine by linking transcriptome or proteomics data. There are few methods based on conventional machine learning algorithms for predicting urine excretory proteins, and most of these methods strongly depend on the extraction of features from urine excretory proteins. An end-to-end model for urine excretory protein prediction, called DeepUEP, is presented using deep neural networks relying on only amino acid sequence information. The model achieves good performance and outperforms existing methods on training and testing sets. By comparing known urinary protein biomarkers with the results of the model, we find that the model can achieve a true-positive rate of over 80% for urinary protein biomarkers that have been detected in more than one study. We also combine our model with transcriptome and proteomics data from lung cancer patients to predict the potential urinary protein biomarkers of lung cancer. A web server is developed for the prediction of urine excretory proteins, and it can be accessed at the following URL: http://www.csbg-jlu.info/DeepUEP/. We believe that our prediction model and web server are useful for biomedical researchers who are interested in identifying urinary protein biomarkers, especially for candidate proteins in transcriptome or proteomics analyses of diseased tissues.

How bacteria respond at the systems level to both genetic and environmental perturbations imposed at the same time is one fundamental yet open question in biology. Bioengineering or synthetic biology provides an ideal system for studying such responses, as engineered strains always have genetic changes as opposed to wildtypes and are grown in conditions which often change during growth for maximal yield of desired products. So, engineered strains were used to address the outstanding question. Two Bacillus subtilis strains (MT1 and MT2) were created previously for the overproduction of N-acetylglucosamine (GlcNAc), which were grown in an environment with a carbon shift from glucose to glucose and xylose in the same culture system. We had four groups: (1) a wildtype (WT) grown with glucose at t1; (2) a WT with glucose and xylose at t2; (3) a mutant (MT1) grown with glucose at t1; and (4) MT1 with glucose and xylose at t2. By measuring transcriptomes and metabolomes, we found that GlcNAc-producing mutants, particularly MT2, had a higher yield of N-acetylglucosamine than WT but displayed a smaller maximum growth rate than the wildtype, despite MT1 reaching higher carrying capacity. Underlying the observed growth, the engineered pathways leading to N-acetylglucosamine had both higher gene expression and associated metabolite concentrations in MT1 than WT at both t1 and t2; in bioenergetics, there was higher energy supply in terms of ATP and GTP, with the energy state metric higher in MT1 than WT at both timepoints. Additionally, most top key precursor metabolites were equally abundant in MT1 and WT at either timepoints. Besides that, one prominent feature was the high consistency between transcriptomics and metabolomics in revealing the response. First, both metabolomes and transcriptomes revealed the same PCA clusters of the four groups. Second, we found that the important functions enriched both by metabolomes and transcriptomes overlapped, such as amino acid metabolism and ABC transport. Strikingly, these functions overlapped those enriched by the genes showing a high (positive or negative) correlation with metabolites. Furthermore, these functions also overlapped the enriched KEGG pathways identified using weighted gene coexpression network analysis. All these findings suggest that the responses to simultaneous genetic and environmental perturbations are well coordinated at the metabolic and transcriptional levels: they rely heavily on bioenergetics, but core metabolism does not differ much, while amino acid metabolism and ABC transport are important. This serves as a design guide for bioengineering, synthetic biology, and systems biology.