how to analyze the co-expression pattern based on Rna-seq DATA

Question

[Jason]

my rna-seq data is from two samples, and the differential expressed genes were available, i want to analyze the differential genes, the relationship of co-expressions, as well as the tissue-specific expressions, any one who has the experience to share with me and many thanks for this wonderful assistance.

[shoujun.gu]

whether the relationship of co-expressions means clustering analysis? and I don't understand why RNAseq can be used for tissue-specific expressions, or maybe your samples came from different tissues?

[Jason]

Yes, you are right, the samples from different tissues, so i want to identify if some genes of tissue-specific expressions. Also, I want to mine more information such as more tools for pathway analysis and co-expression analysis. But some online tools dont matched, such as Genefriend, also Gene Set Enrichment Analysis (GSEA) looks not working.

[prasundutta87]

Check out Miru and its first example dataset: https://kajeka.com/miru/example-data/

[Jason]

Thanks so much for your information. I am trying on this.

Answer 1

[GouthamAtla]

If you have enough samples, you can do a co-expression analysis using WGCNA. This requires a minimum of 15-20 samples as they suggest. With this you may identify group of tightly regulated genes/LncRNAs ( modules ) and hub genes. You can do a functional enrichment analysis on different modules identified by WGCNA. This will help to identify the key regulators ( TF/LncRNA ).

If your data is coming from a specific tissue, you can compare it to bodymap or epigenome roadmap data to identify the tissue specific genes. A k-means clustering approach might work to find the genes expressed exclusively in your samples. You can take all the BAM files and quantify the gene expression levels and normalise the data, and then perform clustering. This should help you to clearly see ubiquitous genes and also blocks of genes expressed in specific tissues.

With differentially expressed genes, you can perform functional enrichment analysis and also some downstream analysis to identify the key regulators using Cytoscape ( Explore the App Store of Cytoscape ) .

You can search on google for specific tools on google.

[Jason]

Thanks so much, this is very helpful.

Answer 2

[theobroma22]

For co-expression as above you can cluster them using the Mfuzz Bioconductor package. You don't state the organism but if you can get the Entrez IDs I recommend to use SPIA Bioconductor package. For tissue specifity you can draw up a model using Limma Bioconductor package. The Biostars community has helped others on these packages so there should be enough resource on this site to get you started and possibly address some FAQs, errors and R code.

[Jason]

THANKS. I will work on this.