Hi there,
I did a de novo structure prediction for a protein (A) and got the PDB file for A. A has another genotype (B). Now I want to predict the structure of B, using A as a template, any website or software can do this?
A and B both are 154 AA long and they have 13 AA difference. I did the de novo prediction for B but when I align the result with A, the difference was quite big, which I think is not accurate. Maybe use A as a template and predict B will be better. I tried Swiss Model but when I try to upload A as a template, it showed there was error. I checked some tutorial, is it the template must be existing in the PDB database but cannot be a predicted one?
Thanks so much!
No, it doesn't have to be a PDB submitted one, most of homology modelling servers would allow you to select your own template lile I-tasser and swiss model. there should be a coordinate error in your pdb file of A, i recommend performing refining process of model A and retry it as a template
you can use
https://zhanglab.ccmb.med.umich.edu/ModRefiner/
http://sysbio.rnet.missouri.edu/3Drefine/index.html
http://csb.stanford.edu/kobamin/
but consider your model A is a homology modeling product so before being able to use it as a template you should estimate it's quality may be you need to do MD simulations not to mention it's sequence identity to it's own template
Hi Tarek,
Thanks for the reply! I use the first link you list to do the refinement and put in Swiss Model as template. It showed the same error like pic 1.I choose user template in pic 3 and click add template file in pic 2.
My protein is virus protein so it's not a homology modeling. It's predicted by QUARK from zhanglab.
Quarks usually get that error of missing coordinates, so please would you try for last
https://zhanglab.ccmb.med.umich.edu/FG-MD/ or http://vienna-ptm.univie.ac.at/ it should add missing paramters, but if not and assuming i get it correct, you have a PDB accession for your"virus protein" yes it happens, we face a lot of high quality structures on PDB database having missing coordinates, so please try this one https://pitgroup.org/decomp/
instead of trying manually using modeler or so, hope that works
Please send me the template.pdb file and sequence of your interested query whoch u want to build 3D structure to my mail id [email protected]
Hi Tarek,
Thanks for the reply! I tried your first link but still cannot. So sad.
Hi,
i-tasser, swiss model, modeller, ... all read the lines above the ATOM statements to find their way in the protein space in orde to predict from your template (depending in how you upload). With a template that was predicted itself however, your pdb files also needs to be perfect.
I am also willing to take a look at it, since i am doing homology modelling and protein prediction daily. I have a colleague here who is also an expert. We use some home made scripts to optimize pdb files (since the smallest error can have detrimental effect)
If you are interested: [email protected]
Good luck.
Hello,
You can try opening A pdb in ucsf chimera and then save a pdb from inside chimera.
It should save a better version of it that you can try for homology modelling.
Please let me know if it doesn't work.
Thank you,
Victor
- Badges
- Science topic