Article
A critical meta-analysis of transcriptomic profiles in neurological tissues during West Nile virus infection
Search Medline for
Authors
-
Klaus Jung
- Stiftung Tierärztliche Hochschule Hannover, Hannover, Deutschland
-
Julien Delarocque
- Stiftung Tierärztliche Hochschule Hannover, Hannover, Deutschland
-
Robin Kosch
- Stiftung Tierärztliche Hochschule Hannover, Hannover, Deutschland
| Published: | August 29, 2017 |
|---|
Outline
Text
Introduction: Infections with the West Nile virus (WNV) often affect neurological tissues also resulting in altered gene expression profiles. In particular, WNV can cause severe neurological diseases in humans, horses and other animals. Several small studies, based on microarray or RNA-seq experiments, have focused on transcriptomic profiles in tissues samples of West Nile infected individuals. Meta-analysis, which is frequently used to combine two or more independent datasets to obtain more precise estimates in a statistical model, can increase the level of scientific evidence. Therefore, we used meta-analysis to reanalyze microarray data from public repositories (GEO, ArrayExpress), with the intention to find differentially expressed genes in infected versus non-infected individuals.
Methods: A systematic data bank search of GEO and ArrayExpress resulted in 48 datasets of WNV-infected mice. Selecting comparable datasets raised several issues. Besides the species of the organism, many more experimental factors exist, which makes it challenging to conduct a meta-analysis. Nevertheless, different study groups were defined, in order to assess the feasibility of meta-analyses with heterogeneous datasets. Particularly, we combined studies with the same neurological tissues, immune cells and immune organs, but differences in specific details (cells, organs, mouse lines and time of sampling post infection). For the analysis the raw data were RMA-preprocessed [1] and merged into a single dataset. Furthermore, batch effects were removed using the ComBat-method [2]. To identify differentially expressed genes, either the limma-Method [3] was used on the merged data, or p-value combinations (MetaMA-method) [4] from the individual studies were performed. All data processing was performed within the R-environment.
Results: Results from the analysis of the merged data and from the p-value combination approach showed a high correlation. The detected genes mainly belong to the immune system, i. e. encode for antiviral functions.
Discussion: Difficulties in the here performed meta-analysis were the small number of available WNV-studies in the databases and the heterogeneity according to their study design. As an outlook, we also plan to employ gene set enrichment analysis in this meta-analysis.
Die Autoren geben an, dass kein Interessenkonflikt besteht.
Die Autoren geben an, dass kein Ethikvotum erforderlich ist.
References
- 1.
- Irizarry RA, Hobbs B, Collin F, Beazer‐Barclay YD, Antonellis KJ, Scherf U, Speed TP. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics. 2003;4(2):249-264.
- 2.
- Leek JT, Johnson WE, Parker HS, Jaffe AE, Storey JD. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics. 2012;28(6):882-883.
- 3.
- Smyth G. Limma: linear models for microarray data. Bioinformatics and computational biology solutions using R and Bioconductor. 2005. p. 397-420.
- 4.
- Marot G, Foulley JL, Mayer CD, Jaffrézic F. Moderated effect size and P-value combinations for microarray meta-analyses. Bioinformatics. 2009; 25(20):2692-2699.
