I am studying the evolutionary aspect of gene family. I want to build the structure of each family member and want to estimate, how much paralogs (family members) are related or distant from each other at structural level.
But I found very low query coverage of target (family member) with template i.e:
Member A: Identity: 30% Query Coverage:49%
Member B: Identity: 33% Query Coverage:49%
Member C: Identity: 31% Query Coverage:52%
Template is available for N terminal of target protein whereas no template is available for C terminal of target protein.
Query coverage and identity of N terminal with target is :
Identity: 35% Query Coverage:91%
I want to Know whether I can build a reliable protein structure for family members using an abinito approach. I am interested in construction of structure for whole protein.
I will be grateful
Dear Anabia,
Your query coverage seems to be very low for predicting the full length protein structure. You said that it is not covering the C-terminal region, which will add the loops in that region. As you said, ab-initio approach may give you the full length structure but it's reliability will need to be confirmed by crystallographic structures and biochemical and biophysical studies. Ab-initio approach is a very time consuming process.
In your case where suitable template is not available for the purpose of predicting the full length structure, you may use the web servers such as I-Tasser (http://zhanglab.ccmb.med.umich.edu/I-TASSER/) which utilizes iterative threading for structure predictions or you may also use Robetta web server (http://robetta.bakerlab.org/submit.jsp) which predict the domains using Ginzu and classifies them into one with available template and one with no template. Then it predicts the structure using both homology modelling and ab-initio modelling.
Once you submit the query in Robetta server, based on your length of protein, it sometimes takes more than one month to return you with the results but as it's approach is based on ab-initio modelling, it may give you the significant results.
Hope it helps.
Best Regards,
a) ProFunc: http://www.ebi.ac.uk/thornton-srv/databases/profunc/
b) Seq2EC (for enzymes only): http://www.ebi.ac.uk/thornton-srv/software/rblec/
using profile-profile alignment for find remote homologous, and use multiple template for modelling. In addition, you can perform secondary prediction using serval servers for have any idea about 3D structure of target and then compare for you model.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs