modeling of large proteins

Question

[gitanjali.tandon]

Hi

I have a protein with 1200 amino acid and I want to model it.

I searched for its templates and I am getting its template till 979 amino acids but how to model the remaining protein.

anyone please help me

Answer 1

[Petr Ponomarenko]

Its tail might not have a stable 3d structure at all. What makes you think there is one? What your model is going to be used for?

Answer 2

[gitanjali.tandon]

Thanks for your reply.. I have to use this model for further studying protein - protein interaction.

[Petr Ponomarenko]

Do you have any evidence that aminoacids in that tail participate in the interaction with other protein? For protein protein interaction you need to model interacting area very well first in order to find key aminoacids and their important properties for that interaction. Than you build empirical model of interaction based on this. If you have a list of such aminoacids and can gues their important properties you can skip 3D structure modeling or only model these aminoacids interaction. When there is very little information on aminoacids interaction I usually start with 3D structures of similar protein pairs that include interaction already. Than I use these structures together with aporopriate templates for homology modeling and interaction modeling. After I have interaction model I doublecheck that it makes sense from experimental data like directed mutagenesis. When I get such validated model I search for key aminoacids in it and refine structure manually around them. Than I create an empirical function with unknown parameters based on several properties of these key aminoacids potentially important for function that these binding plays. Programs tent to use energy calculation that was optimized for certain tasks, like protein structure refining, molecular docking, etc. For functional predictions completely different aspects can be important like size, charge or ability to form sulfur bond of one particular aminoacid. Finally I optimize unknown parameters in my function based on experimental data of interaction. I find this approach has more statistical power. The reason is in the way energy functions are optimized in different programs.

[gitanjali.tandon]

Thanks a lot for your reply.. Actually I don't have any information of which part of this protein is interacting with another protein.. This is what I wanted to find out.. So I first proceeded with identifying the important domains of my proteins keeping in view that these domains are modelled so that interaction can be studied properly.. Is there any way out that we model the different parts of the protein and then combining them.. If yes can you please assist me in this..

[Petr Ponomarenko]

You have to have at least some experimental information about this protein pair interaction. Are they binding as monomers or dimers? What parts of these proteins are important for the binding? Are there any mutations or drugs affecting that binding? Maybe you know from 3d structures that one of this proteins binds to some other (third) protein. Do you know experimentally that these affects your protein pair interaction? Is one of the proteins a membrane protein? Do you know from experiments that one of the proteins in a pair interact with a truncated version of the second protein in a pair?

You will have to address at least some of these questions in your publication to confirm your model anyway, so digging deep into the literature about these proteins is the first step. Understanding what experiments can be done to prove your model is also critical as sometimes you will have to make them before the publication simply because prior literature and your modeling is not enough to argue your ideas some times.

[WouterDeCoster]

Please use ADD COMMENT to answer to earlier replies, as such this thread remains logically structured and easy to follow.