Are There Any Differences Between Human Protein-Protein Interaction And Mouse Protien-Protien Interaction?

Question

[michaelchen333333]

In my project, I need to use "human" embryonic stem cell related protein-protein interaction data, but I only find a few human ESC-related PPI data in BioGRID, HPRD, InAct. In contrast, there is a lot of mouse ESC-related PPI data in these databases.

I know that human genome is about 90% similar to mouse genome, so, is it ok if I combine the mouse PPI data with human PPI data to use? (I will use NCBI HomoloGene to map mouse gene to human gene)

Any suggestion will be nice.

Answer 1

[pld]

Yes, there are major differences. Coding similarity is not the same as regulatory similarity: just because there is an ortholog, it doesn't mean that gene behaves exactly the same in the two species.

Mice are good models because they are close to humans, they are cheap to purchase, require simple husbandry, and aren't going to give you B virus. Best of all you can get a few hundred mice in each study group without issue. You also have to remember that model organisms aren't chosen for complete biology, but being a good phenotypic approximation to whatever aspect of human biology people wish to test. They're awseome for some things, and useless for others.

Given how important developmental biology is for cancer research, I am surprised that you're having trouble finding human models and PPI data for the developmental pathways you're interested in. I think a better approach would be to generate a list of proteins and gene products you're interested in and then mine all available PPI data, paying close attention to cancer stuff. This is an important lesson I've learned about mammalian systems biology: it is almost all cancer stuff. If I want to look at immune system stuff for viral infections, I won't find anything just searching for that, I have to go and dig what I want out of cancer data.

In your case, take all of the genes/pathways in the mouse data, and then pull interactions for those genes/etc from any human PPI data you can get your hands on.

If for some reason you can only get what you want from mouse data, I would cross check your major regulatory sites. Phosphorylation, Histone binding/modification sites, methylation and transcriptional regulation. Phosphorylation might be tough, but for the others there is loads of ChIP-Seq data out there. It will at least give you some idea of how close of an approximation you're working with.

[michaelchen333333]

Thanks, good points, I will recheck my approach.

Answer 2

[DG]

There will be probably be differences in biological reality. If you look at some of the ortholog/paralog papers from a few years ago about functional differences, you do see things like differences in human/mouse tissue expression patterns for orthologs versus paralogs for instance. That said the Mouse data is still a good starting point in terms of doing data mining and generating hypotheses that can be followed up on with wet lab experiments in actual human ESC lines.

[michaelchen333333]

Thanks, I will think carefully about using the mouse PPI data.