In silico protein modeling

Question

[ravihansa82]

Hi all, I am currently working in protein - protein interaction. One of my objectives was to identify the interaction domains of those two proteins. Luckily I found the interaction domains. One of those two proteins is a novel protein. There was no homology sequence found so far. Interestingly, that protein is predicted to be a intrinsically disordered protein. The other protein is an RNA helicase. However, I could purify the RNA helicase protein interacting Domain in disordered protein and the RNA helicase protein for the activity assay. At the moment I am trying to crystallize those purified Domain from disordered protein with RNA helicase. Unfortunately, seems it takes considerable amount of optimization to get the crystals. Since, I have short time to finish my project I need to find out a way to determine, amino acids which involves for this protein - protein interaction. I got to know there is a possibility to model the protein - protein interaction in silico. I already found a model for the RNA helicase based on homology modeling through I-TESSER SETVER. however, I could not find such model for the other protein. I really appreciate if you can share your ideas with me to model this protein interaction to find out the amino acids in the interface of the two proteins. If it is possible with in silico approach, I really appreciate your kind suggestions and guidance. Thanks

Answer 1

[Mensur Dlakic]

Not going to sugarcoat it for you - based on what you describes, I doubt that you will be able to get meaningful information from in silico analysis. First, it seems like you don't have structures for either protein. Next level below that would be having good homology models of both proteins. It seems like you may have one, but it really doesn't help that your other protein is disordered. Docking two structured proteins solved at high resolution would not be a trivial task given that you lack information about their potential interaction interface. Doing it with a homology model and a suspect model of an intrinsically disordered protein sounds like a non-starter to me.

Not to be all gloomy, but unless your disordered proteins get structured upon interaction - it has been known to happen - it doesn't bode well for crystallizing the complex either. Have you tried predicting the interaction surface based on co-variation analysis?

https://elifesciences.org/articles/03430

http://evcouplings.org/

https://www.nature.com/articles/srep30425

https://academic.oup.com/nar/article/46/W1/W323/5037718

https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-016-0975-z