I performed a differential expression analysis and I have a list of DE human genes. now, I want to visualize DEGs by Cytoscape (showing up-regulated and down-regulated proteins with different shapes and set the size of the nods on the base of levels of expression.) but this work needs a network and expression data.how should I obtain these files? And after that how to connect them to build a network?
Article Biological Network Exploration with Cytoscape 3
Article Cytoscape: The network visualization tool for GenomeSpace workflows
Article Enrichment Map: A Network-Based Method for Gene-Set Enrichme...
Hi Maryam Momeni ,
There are numerous ways to deal with this kind of data, but it depends on your primary goal. If you want to build a network for your DE genes, you first need a network file providing your nodes/edges information.
A straightforward way to achieve this is by using the STRING engine [https://string-db.org/]. Just plot your input there, and create a primary network. Some genes might not connect, but you can mess around with the search parameters to suit your needs.
After you have this primary network, you can create an attribute table using your DE and import it to Cytoscape as a Node Attribute. This way, you will have your system together with the expression information.
A more annoying way is to build the input file yourself. However, you will need to define by yourself the edge's information. This step is modulated by the type of data you have and what do you want to analyze. For example, if you wish to a directed network or an undirected network. Additionally, you can decide if they are based on repression, activation, or some score you calculated elsewhere.
I hope this helps.
You can construct a gene co-expression network by calculating the Pearson correlation coefficient using gene expression values.
After that export, the correlation matrix in the Cytoscape and visualize it, you can also create an attribute table using your DE and import it to Cytoscape as a Node Attribute.
Rinki Singh
Thanks for answering my question. I would be more appreciated if you explain to me how to construct such a co-expression network? By Which tools?
Bruno César Feltes
Hi. Thanks for your answer. You stated, "You can mess around with the search parameters to suit your needs". I would be more appreciated if you explain what can I do with the gene symbols that are known by STRING for more than one protein? which search parameters do you mean?
My other question is How can I create an attribute table? Which columns it should have? And How can I enter it to Cytoscape?
ExpressionCorrelation is a Cytoscape’s plugin provide a similarity matrix computed by pearson correlation. In addition to being user-friendly, the main advantage of this tool is that the reconstructed networks are directly imported in Cytoscape where it may be evaluated by other plugins.
WGCNA is another most used approach to build and to analyze gene co-expression networks
Rinki Singh
Now I have a list of gene symbols and logFC s. How can I prepare an input for ExpressionCorrelation plugin of cytoscape?
Maryam Momeni,
Hello. Sure, I will explain. When a gene symbol retrieves more than one gene, you will need to select the gene manually. It isn't very pleasant, but it is better just blindly to trust the first option. The parameters I mentioned can be found on the "Setting" tab after you have the network display. You can read the tutorials and explanations on the STRING website because they are straightforward and simple to understand. In a nutshell, you can set the type of interaction, the confidence score, and add nodes to your network from there. Test different parameters combination to suit your needs (just don`t add textmining, not trustworthy).
Attribute tables are flexible and can be molded by your needs. The columns should state your seed gene, and the gene should be connected, as well, for the column that represents the interaction. It doesn't matter which column is each because you can manually select each one when you import a table in Cytoscape.
Now, I must advise you to think about what do you really want to analyze here. The method I am explaining is about visualizing them in a "network view" so you can make topological analyzes after (e.g., clustering, degree, betweenness, and so on). If you want to make a co-expression network, that's another story. It is way more complex, and it will provide a different type of result. At this point, it truly depends on what do you want to analyze.
Cheers.
Hello Momeni, I got what you want, you follow this questions and answers. It will help you.
https://www.biostars.org/p/192885/
- Badges
- Science topic
- Similar topics
- Biological Science
- Bioecology
- Systems Ecology
- Landscape Ecology
- Connectivity