Meta-analysis method for discovering reliable biomarkers by integrating statistical and biological approaches: An application to liver toxicity

Integrative OMICS Data-Driven Procedure Using a Derivatized Meta-Analysis Approach

The wealth of high-throughput data has opened up new opportunities to analyze and describe biological processes at higher resolution, ultimately leading to a significant acceleration of scientific output using high-throughput data from the different omics layers and the generation of databases to store and report raw datasets. The great variability among the techniques and the heterogeneous methodologies used to produce this data have placed meta-analysis methods as one of the approaches of choice to correlate the resultant large-scale datasets from different research groups. Through multi-study meta-analyses, it is possible to generate results with greater statistical power compared to individual analyses. Gene signatures, biomarkers and pathways that provide new insights of a phenotype of interest have been identified by the analysis of large-scale datasets in several fields of science. However, despite all the efforts, a standardized regulation to report large-scale data and to identify the molecular targets and signaling networks is still lacking. Integrative analyses have also been introduced as complementation and augmentation for meta-analysis methodologies to generate novel hypotheses. Currently, there is no universal method established and the different methods available follow different purposes. Herein we describe a new unifying, scalable and straightforward methodology to meta-analyze different omics outputs, but also to integrate the significant outcomes into novel pathways describing biological processes of interest. The significance of using proper molecular identifiers is highlighted as well as the potential to further correlate molecules from different regulatory levels. To show the methodology’s potential, a set of transcriptomic datasets are meta-analyzed as an example.

Role of Transportome in the Gills of Chinese Mitten Crabs in Response to Salinity Change: A Meta-Analysis of RNA-Seq Datasets

Chinese mitten crab (CMC) or Eriocheir sinensis is a strong osmoregulator that can keep rigorous cellular homeostasis. CMC can flourish in freshwater, as well as seawater, habitats and represents the most important species for freshwater aquaculture. Salt stress can have direct effects on several stages (e.g., reproduction, molting, growth, etc.) of the CMC life cycle. To get a better overview of the genes involved in the gills of CMC under different salinity conditions, we conducted an RNA-Seq meta-analysis on the transcriptomes of four publicly available datasets. The meta-analysis identified 405 differentially expressed transcripts (DETs), of which 40% were classified into various transporter classes, including accessory factors and primary active transporters as the major transport classes. A network analysis of the DETs revealed that adaptation to salinity is a highly regulated mechanism in which different functional modules play essential roles. To the best of our knowledge, this study is the first to conduct a transcriptome meta-analysis of gills from crab RNA-Seq datasets under salinity. Additionally, this study is also the first to focus on the differential expression of diverse transporters and channels (transportome) in CMC. Our meta-analysis opens new avenues for a better understanding of the osmoregulation mechanism and the selection of potential transporters associated with salinity change.

Integrated meta-analysis and machine learning approach identifies acyl-CoA thioesterase with other novel genes responsible for biofilm development in Staphylococcus aureus

Biofilm forming Staphylococcus aureus is a major threat to the health-care industry. It is important to understand the differences between planktonic and biofilm growth forms in the pathogen since conventional treatments targeting the planktonic forms are not effective against biofilms. The current study conducts a meta-analysis of three public transcriptomic profiles to examine the differences in gene expression between the planktonic and biofilm states of S. aureus using random-effects modeling. Mean effect sizes were calculated for 2847 genes among which 726 differentially expressed genes were taken for further analysis. Major genes that are discriminatory between the two conditions were mined using supervised learning techniques and validated by high-accuracy classifiers. Ten different feature selection algorithms were applied and used to rank the most important genes in S. aureus biofilms. Finally, an optimal set of 36 genes are presented as candidate genes in biofilm formation or development while throwing light on the novel roles of an acyl-CoA thioesterase enzyme and 10 hypothetical proteins in biofilms. The relevance of the identified gene set was further validated by building five different classification models using SVM, RF, kNN, NB and DT algorithms that were compared with models built from other relevant gene sets and by reviewing the functional role of 25 previously known genes in biofilm development. The study combines meta-analysis of differential expression with supervised machine learning strategies and feature selection for the first time to identify and validate a discriminatory set of genes important in biofilms of S. aureus. The functional roles of the identified genes predicted to be important in biofilms are further scrutinized and can be considered as a signature target list to develop anti-biofilm therapeutics in S. aureus.

Validating candidate biomarkers for different stages of non-alcoholic fatty liver disease

Non-alcoholic fatty liver disease (NAFLD) is a common chronic condition caused by the accumulation of fat in the liver. NAFLD may range from simple steatosis to advanced cirrhosis, and affects more than 1 billion people around the world. To date, there has been no effective treatment for NAFLD. In this study, we evaluated the expression of 4 candidate NAFLD biomarkers to assess their possible applicability in the classification and treatment of the disease.Twenty-six obese subjects, who underwent bariatric surgery, were recruited and their liver biopsies obtained. Expression of 4 candidate biomarker genes, PNPLA3, COL1A1, PPP1R3B, and KLF6 were evaluated at gene and protein levels by RT-qPCR and enzyme-linked immunosorbent assay (ELISA), respectively.A significant increase in the levels of COL1A1 protein (P = .03) and PNPLA3 protein (P = .03) were observed in patients with fibrosis-stage NAFLD compared to that in patients with steatosis-stage NAFLD. However, no significant differences were found in abundance of PPP1R3B and KLF6 proteins or at the gene level for any of the candidate.This is the first study, to our knowledge, to report on the expression levels of candidate biomarker genes for NAFLD in the Saudi population. Although PNPLA3 and PPP1R3B had been previously suggested as biomarkers for steatosis and KLF6 as a possible marker for the fibrosis stage of NAFLD, our results did not support these findings. However, other studies that had linked PNPLA3 to fibrosis in advanced NAFLD supported our current finding of high PNPLA3 protein in patients with fibrosis. Additionally, our results support COL1A1 protein as a potential biomarker for the fibrosis stage of NAFLD, and indicate its use in the screening of patients with NAFLD. Further studies are required to validate the use of COL1A1 as a biomarker for advanced NAFLD in a larger cohort.

A survey of gene expression meta-analysis: methods and applications

The increasing use of high-throughput gene expression quantification technologies over the last two decades and the fact that most of the published studies are stored in public databases has triggered an explosion of studies available through public repositories. All this information offers an invaluable resource for reuse to generate new knowledge and scientific findings. In this context, great interest has been focused on meta-analysis methods to integrate and jointly analyze different gene expression datasets. In this work, we describe the main steps in the gene expression meta-analysis, from data preparation to the state-of-the art statistical methods. We also analyze the main types of applications and problems that can be approached in gene expression meta-analysis studies and provide a comparative overview of the available software and bioinformatics tools. Moreover, a practical guide for choosing the most appropriate method in each case is also provided.

Identification of key genes and long non-coding RNA associated ceRNA networks in hepatocellular carcinoma

Background

Hepatocellular carcinoma (HCC) is one of the leading causes of cancer-related deaths worldwide. Although multiple efforts have been made to understand the development of HCC, morbidity, and mortality rates remain high. In this study, we aimed to discover the mRNAs and long non-coding RNAs (lncRNAs) that contribute to the progression of HCC. We constructed a lncRNA-related competitive endogenous RNA (ceRNA) network to elucidate the molecular regulatory mechanism underlying HCC.

Methods

A microarray dataset (GSE54238) containing information about both mRNAs and lncRNAs was downloaded from the Gene Expression Omnibus database. Differentially expressed genes (DEGs) and lncRNAs (DElncRNAs) in tumor tissues and non-cancerous tissues were identified using the limma package of the R software. The miRNAs that are targeted by DElncRNAs were predicted using miRcode, while the target mRNAs of miRNAs were retrieved from miRDB, miRTarBas, and TargetScan. Functional annotation and pathway enrichment of DEGs were performed using the EnrichNet website. We constructed a protein–protein interaction (PPI) network of DEGs using STRING, and identified the hub genes using Cytoscape. Survival analysis of the hub genes and DElncRNAs was performed using the gene expression profiling interactive analysis database. The expression of molecules with prognostic values was validated on the UALCAN database. The hepatic expression of hub genes was examined using the Human Protein Atlas. The hub genes and DElncRNAs with prognostic values as well as the predictive miRNAs were selected to construct the ceRNA networks.

Results

We found that 10 hub genes (KPNA2, MCM7, CKS2, KIF23, HMGB2, ZWINT, E2F1, MCM4, H2AFX, and EZH2) and four lncRNAs (FAM182B, SNHG6, SNHG1, and SNHG3) with prognostic values were overexpressed in the hepatic tumor samples. We also constructed a network containing 10 lncRNA–miRNA–mRNA pathways, which might be responsible for regulating the biological mechanisms underlying HCC.

Conclusion

We found that the 10 significantly overexpressed hub genes and four lncRNAs were negatively correlated with the prognosis of HCC. Further, we suggest that lncRNA SNHG1 and the SNHG3-related ceRNAs can be potential research targets for exploring the molecular mechanisms of HCC.

Impacts of age and gender at the risk of underlying medical conditions and death in patients with avian influenza A (H7N9): a meta-analysis study

Objective

The objective of our study was to conduct a series of analyses that examined the impacts of age and gender at the risk of underlying medical conditions (UMCs) and death in patients with influenza A (H7N9).

Methods

We began by searching for potentially relevant articles in English or Chinese before February 28, 2018. Additionally, we reviewed our own files and reference lists of articles identified by this search.

Results

The association between death and UMCs was significant in H7N9 patients, with an OR of 1.49 (95% CI: 1.24–1.78). Subgroup analyses showed that having two or more UMCs of any type (OR: 2.24; P=0.044), chronic respiratory diseases (OR: 1.81; P=0.032), and chronic cardiovascular disease (OR: 1.63; P=0.013) had an association with increased fatality in H7N9 patients. Age (60 years or older) [adjusted OR (AOR): 1.86; P=0.032] and gender (male: AOR: 1.68, P=0.006; female: AOR: 1.88, P=0.044) were significantly associated with death in H7N9 patients with UMCs compared to H7N9 patients without any UMC. Stratification analyses found statistically significant increased death in H7N9 patients with UMCs who were 60 years of age and older (AOR: 2.72; P<0.001) and gender (male; AOR=1.64; P=0.033), compared to H7N9 patients without these respective conditions.

Conclusion

Impacts of age are substantial and significant at the risk of UMCs and death in H7N9 patients. This analysis did not find a significant difference in gender comparisons. Efforts should particularly focus on reducing fatality rates in patients with combined risks from UMCs and other significant impact factor such as age (60 years or older).