Expression Network vs PPI Network

Question

[shahroze786]

I have a general question regarding the type of information that can be provided by different types of networks. What are the advantages of constructing an expression network for analysis compared to a PPI network? What will one give more in regards to studying a disease for example? If I use basal brain tissue gene expression to construct a network, compared to using PPI from STRING, will one provide more relevant info than the other? I know that a gene will encode for a specific protein but just because geneA is highly correlated with geneB, doesn't necessarily mean their protein products will have high correlation. Thanks in advance.

Answer 1

[Jean-Karim Heriche]

It depends on what information is relevant to you. Basically, for mammalian systems, expression data are good for fingerprinting while protein interactions give you information on gene function. So expression can be useful to assess sample similarities while protein interaction networks allow to predict gene function. However, data of any kind is useless if you don't have a biological question.

Answer 2

[fishgolden]

If I use basal brain tissue gene expression to construct a network, compared to using PPI from STRING, will one provide more relevant info than the other? I know that a gene will encode for a specific protein but just because geneA is highly correlated with geneB, doesn't necessarily mean their protein products will have high correlation.

Yes, PPI network gives much more direct information than expression network (at least you focus on works done by coded proteins).

However, getting data of PPI or calculating amount of proteins in cells are more difficult or expensive than getting expression data. Therefore expression network is a good source to analyze.

[shahroze786]

Do you know where I could download mouse brain PPI data? I've found sources online but most seem to just provide a web interface to interact with the API...

[fishgolden]

No. I have used intact https://www.ebi.ac.uk/intact/downloads , but it is not tissue specific, I think.

[Jean-Karim Heriche]

One way to deal with the absence of tissue-specific PPI data is to filter the interactions to only keep those in which both partners are expressed in the tissue of interest. This is what the TissueNet database does (for human proteins though).