Homology Modelling - (Apr/16/2007 )
Can any body tell me when searching for template for homology modelling what is criteria for choosing best templates.
One have to look for identity and similarity values for the sequence.
you mean protein homology, it's mostly about amino acid esquence
Yeah protein modelling. Similaitry and identity are two different things. What one have to look. Either identity or similarity. What about gap values.
If Identity is low and gap is more. Which template will be good with high identity or low gap or high gap with low identity.
If you want to do comparative then you should look for a sequence with known structure that shares a maximum sequence register with few gaps and high sequence identity.
You want to get a homolog. Typically a sequence with identity to the template (sequence with known structure). The lower you drop in sequence ID the close you get to threading!!!
I would recommed the SWISS-MODEL server at here http://swissmodel.expasy.org//SWISS-MODEL.html, their software can search for homology modeling template automatically.
I agree, it is quite a good software, Cn3D is also good.
But I remember that you need at least 80% similarity to do modeling.
Yeah this is differnce concept similatry and homology is two different concepts. I need high similairy or high homology. I donot think one is able to get 80 percent homology. i think more the 35 percent is enough to select pdb model as a template for homology modelling. what you think
I agree, it is quite a good software, Cn3D is also good.
But I remember that you need at least 80% similarity to do modeling.
One have to look for identity and similarity values for the sequence.
Hi,
Others have just say to you some properties of identity and similarity, I don't known your protein properties but first you can use more programs for "define" this properties...
maybe you can split your protein in stuctural domain or function domain and after or combined found your template for homology modelling.
Bye.
M.
Hey,
just to be a pedant, two sequences are homologous or not. You don't have levels of homology.
You can do comparative modeling at 35% sequence similarity, however I would suggest that you didn't do this if there are more similar structures out there. If you want a structure, you would be wise to contact an expert.
I am not sure if the Baker lab has their software online but it would be worth a look.