Friday at 11:57 am
- Link
Thomas Lemberger, joergkurtwegner, Timothy Driscoll and 6 other people liked this
We have over 1100 HCV-Hu ppi in PIG, whereas the authors have far fewer. Not sure why. (PIG aggrepates ppi from public databases like bind and such.) - Timothy Driscoll
via Alert Thingy
There have been so many recent -omics studies of host-pathogen interactions (both physical and functional) that it would be very worth while to attempt some integration/comparative studies. One thing that is typically seen in these studies is that pathogen proteins interact preferentially with host proteins that are already highly connected. They have still to offer some biological explanation as to why this happens. - Pedro Beltrao
Timothy - do you or any of your sources transfer interactions by homology ? - Pedro Beltrao
Not to my knowledge; they are all expeirmentally confirmed interactions in PIG. - Timothy Driscoll
via Alert Thingy
ack; PIGs interactions are a non-rendundant set. never mind. :-) - Timothy Driscoll
via Alert Thingy
Then again, virus and host first interact via extra-cellular receptors, which are amongst the fastest evolving regimes in the eukaryotic cell and badly assessed in Y2H-screens. Regarding pathogen proteins and their degree - did any screen ever reveal a class of proteins that preferentially interacts with less connected proteins? - Roland Krause
Roland - That is a good point. I guess one possible test would be to take a similar number of random host proteins (as the pathogen proteins tested) and see how they interact with the other host proteins. This would provide with a random model to test if the pathogen-host interaction properties are significantly different from any random interaction screen. - Pedro Beltrao
The randomization model needs to capture more than just the degree distribution - but I cannot come up with a complete list of how to address the problem as of now. All screens, no matter what is studied connect to proteins with higher degree and than expected. - Roland Krause



