Haixiu Yang

Biomedical ontologies, such as Gene Ontology (GO), Disease Ontology (DO), and the Human Phenotype Ontology (HPO), have been extensively applied to characterize molecular roles and their semantic relationships in biomedical research and clinical practice. Although numerous algorithms have been developed to quantify relationships between ontology terms or to explore molecular functions, the absence of a comprehensive tool to integrate these algorithms has limited effective ontology applications. To address this, we developed OntoTiger, a platform of Ontology-based application Tools for InteGrativE biomedical exploRation. OntoTiger combines >20 classic algorithms, supporting six prevalent molecular types as well as five widespread biomedical ontologies. The platform comprises four modules: (i) Annotation module, which qualifies the relationships between ontology terms and molecules; (ii) Similarity module, quantifying functional similarity between/across pairwise ontology terms or between molecules; (iii) Prediction module, characterizing the molecular roles from an ontological perspective; and (iv) Enrichment module, elucidating the potential biological significance of a particular list of molecules. OntoTiger provides a freely accessible, user-friendly web server dedicated to enabling one-stop ontology-based applications and is freely available at https://bio-computing.hrbmu.edu.cn/OntoTiger.

Disease Ontology (DO) has been widely studied in biomedical research and clinical practice to describe the roles of genes. DO enrichment analysis is an effective means to discover associations between genes and diseases. Compared to hundreds of Gene Ontology (GO)–based enrichment analysis methods, however, DO-based methods are relatively scarce, and most current DO-based approaches are term-for-term and thus are unable to solve over-enrichment problems caused by the “true-path” rule. Here, we describe a novel double-weighted model, EnrichDO, which leverages the latest annotations of the human genome with DO terms and integrates DO graph topology on a global scale. Compared to classic enrichment methods (mainly for GO) and existing DO-based enrichment tools, EnrichDO performs better in both GO and DO enrichment analysis cases. It can accurately identify more specific terms, without ignoring the truly associated parent terms, as shown in the Alzheimer’s disease (AD) case (AD ranked first). Moreover, both a simulated test and a data perturbation test validate the accuracy and robustness of EnrichDO. Finally, EnrichDO is applied to other types of datasets to expand its application, including gene expression profile datasets, a host gene set of microorganisms, and hallmark gene sets. Based on the findings reported here, EnrichDO shows significant improvement via all experimental results. EnrichDO provides an effective DO enrichment analysis model for gaining insight into the significance of a particular gene set in the context of disease. To increase the usability of EnrichDO, we have developed an R-based software package, which is freely available through Bioconductor (https://bioconductor.org/packages/release/bioc/html/EnrichDO.html) or at https://github.com/liangcheng-hrbmu/EnrichDO.

Biomedical big data, characterized by its massive scale, multi-dimensionality, and heterogeneity, offers novel perspectives for disease research, elucidates biological principles, and simultaneously prompts changes in related research methodologies. Biomedical ontology, as a shared formal conceptual system, not only offers standardized terms for multi-source biomedical data but also provides a solid data foundation and framework for biomedical research. In this review, we summarize enrichment analysis and deep learning for biomedical ontology based on its structure and semantic annotation properties, highlighting how technological advancements are enabling the more comprehensive use of ontology information. Enrichment analysis represents an important application of ontology to elucidate the potential biological significance for a particular molecular list. Deep learning, on the other hand, represents an increasingly powerful analytical tool that can be more widely combined with ontology for analysis and prediction. With the continuous evolution of big data technologies, the integration of these technologies with biomedical ontologies is opening up exciting new possibilities for advancing biomedical research.

Ischemic cardiomyopathy (ICM) and dilated cardiomyopathy (DCM) are the two main causes of heart failure (HF). Despite similar clinical characteristics and common “HF pathways”, ICM and DCM are expected to have different personalized treatment strategies. The underlying mechanisms of ICM and DCM have yet to be fully elucidated. The present study developed a novel computational method for identifying dysregulated long noncoding RNA (lncRNA)–microRNA (miRNA)–mRNA competing endogenous RNA (ceRNA) triplets. Time-ordered dysregulated ceRNA networks were subsequently constructed to reveal the possible disease progression of ICM and DCM based on the method. Biological functional analysis indicated that ICM and DCM had similar features during myocardial remodeling, whereas their characteristics differed during progression. Specifically, disturbance of myocardial energy metabolism may be the main characteristic during DCM progression, whereas early inflammation and response to oxygen are the characteristics that may be specific to ICM. In addition, several panels of diagnostic biomarkers for differentiating non-heart failure (NF) and ICM (NF-ICM), NF-DCM, and ICM-DCM were identified. Our study reveals biological differences during ICM and DCM progression and provides potential diagnostic biomarkers for ICM and DCM.

Aberrant metabolism is one of the main driving forces in the initiation and development of ESCC. Both genes and metabolites play important roles in metabolic pathways. Integrative pathway analysis of both genes and metabolites will thus help to interpret the underlying biological phenomena. Here, we performed integrative pathway analysis of gene and metabolite profiles by analyzing six gene expression profiles and seven metabolite profiles of ESCC. Multiple known and novel subpathways associated with ESCC, such as ‘beta-Alanine metabolism’, were identified via the cooperative use of differential genes, differential metabolites, and their positional importance information in pathways. Furthermore, a global ESCC-Related Metabolic (ERM) network was constructed and 31 modules were identified on the basis of clustering analysis in the ERM network. We found that the three modules located just to the center regions of the ERM network—especially the core region of Module_1—primarily consisted of aldehyde dehydrogenase (ALDH) superfamily members, which contributes to the development of ESCC. For Module_4, pyruvate and the genes and metabolites in its adjacent region were clustered together, and formed a core region within the module. Several prognostic genes, including GPT, ALDH1B1, ABAT, WBSCR22 and MDH1, appeared in the three center modules of the network, suggesting that they can become potentially prognostic markers in ESCC.

Dysregulated long non-coding RNAs (lncRNAs) have been found to have oncogenic and/or tumor suppressive roles in the development and progression of cancer, implying their potentials as novel independent biomarkers for cancer diagnosis and prognosis. However, the prognostic significance of expression profile-based lncRNA signature for outcome prediction in patients with multiple myeloma (MM) has not yet been investigated. LncRNA expression profiles of a large cohort of patients with MM were obtained and analyzed by repurposing the publically available microarray data. An lncRNA-focus risk score model was developed from the training dataset, and then validated in the testing and another two independent external datasets. The time-dependent receiver operating characteristic (ROC) curve was used to evaluate the prognostic performance for survival prediction. The biological function of prognostic lncRNAs was predicted using bioinformatics analysis. Four lncRNAs were identified to be significantly associated with overall survival (OS) of patients with MM in the training dataset, and were combined to develop a four-lncRNA prognostic signature to stratify patients into high-risk and low-risk groups. Patients of training dataset in the high-risk group exhibited shorter OS than those in the low-risk group (HR = 2.718, 95 % CI = 1.937-3.815, p <0.001). The similar prognostic values of four-lncRNA signature were observed in the testing dataset, entire GSE24080 dataset and another two independent external datasets. Multivariate Cox regression and stratified analysis showed that the prognostic power of four-lncRNA signature was independent of clinical features, including serum beta 2-microglobulin (Sβ2M), serum albumin (ALB) and lactate dehydrogenase (LDH). ROC analysis also demonstrated the better performance for predicting 3-year OS. Functional enrichment analysis suggested that these four lncRNAs may be involved in known genetic and epigenetic events linked to MM. Our results demonstrated potential application of lncRNAs as novel independent biomarkers for diagnosis and prognosis in MM. These lncRNA biomarkers may contribute to the understanding of underlying molecular basis of MM.

Accumulated evidence suggests that dysregulated expression of long non-coding RNAs (lncRNAs) may play a critical role in tumorigenesis and prognosis of cancer, indicating the potential utility of lncRNAs as cancer prognostic or diagnostic markers. However, the power of lncRNA signatures in predicting the survival of patients with non-small cell lung cancer (NSCLC) has not yet been investigated. We performed an array-based transcriptional analysis of lncRNAs in large patient cohorts with NSCLC by repurposing microarray probes from the gene expression omnibus database. A risk score model was constructed based on the expression data of these eight lncRNAs in the training dataset of NSCLC patients and was subsequently validated in other two independent NSCLC datasets. The biological implications of prognostic lncRNAs were also analyzed using the functional enrichment analysis. An expression pattern of eight lncRNAs was found to be significantly associated with overall survival (OS) of NSCLC patients in the training dataset. With the eight-lncRNA signature, patients of the training dataset could be classified into high- and low-risk groups with significantly different OS (median survival 1.67 vs. 6.06 years, log-rank test p = 4.33E−09). The prognostic power of eight-lncRNA signature was further validated in other two non-overlapping independent NSCLC cohorts, demonstrating good reproducibility and robustness of this eight-lncRNA signature in predicting OS of NSCLC patients. Multivariate regression and stratified analysis suggested that the prognostic power of the eight-lncRNA signature was independent of clinical and pathological factors. Functional enrichment analyses revealed potential functional roles of the eight prognostic lncRNAs in tumorigenesis. These findings indicate that the eight-lncRNA signature may be an effective independent prognostic molecular biomarker in the prediction of NSCLC patient survival.

Identification of key metabolites for complex diseases is a challenging task in today's medicine and biology. A special disease is usually caused by the alteration of a series of functional related metabolites having a global influence on the metabolic network. Moreover, the metabolites in the same metabolic pathway are often associated with the same or similar disease. Based on these functional relationships between metabolites in the context of metabolic pathways, we here presented a pathway-based random walk method called PROFANCY for prioritization of candidate disease metabolites. Our strategy not only takes advantage of the global functional relationships between metabolites but also sufficiently exploits the functionally modular nature of metabolic networks. Our approach proved successful in prioritizing known metabolites for 71 diseases with an AUC value of 0.895. We also assessed the performance of PROFANCY on 16 disease classes and found that 4 classes achieved an AUC value over 0.95. To investigate the robustness of the PROFANCY, we repeated all the analyses in two metabolic networks and obtained similar results. Then we applied our approach to Alzheimer's disease (AD) and found that a top ranked candidate was potentially related to AD but had not been reported previously. Furthermore, our method was applicable to prioritize the metabolites from metabolomic profiles of prostate cancer. The PROFANCY could identify prostate cancer related-metabolites that are supported by literatures but not considered to be significantly differential by traditional differential analysis. We also developed a freely accessible web-based and R-based tool at http://bioinfo.hrbmu.edu.cn/PROFANCY.