I think none of tertiary structure prediction method can predict structure with high accuracy as homology is very poor. In this situation I will advise you to predict secondary structure information using following software phsipred, betatpred2, alphapred etc.

Dear Shivakrishna. Just give a try to check if the model you obtain could be reliable or not. Whether sequence identity is important in modelling, and of course, a high sequence identity will support the robustness of your model, low sequence identity models are also reliable if the fold of your protein is conserved (e.g., if they belong to the same family of proteins)

Try, HHPred or HHsuite

You can also see http://expasy.org/ and related sub-pages where a lot of interesting softare are reported (i.e., http://swissmodel.expasy.org/)

Apart from sequence identity of target-template you should also consider the query coverage of the target. If query coverage is >80% then u can go ahead ...

It is only 18 %

I think none of tertiary structure prediction method can predict structure with high accuracy as homology is very poor. In this situation I will advise you to predict secondary structure information using following software phsipred, betatpred2, alphapred etc.

I agree with Prof. Raghava as sequence similarity is very low. On the other hand, if you are lucky, some of the structural modelling softwares mentioned above can handle this situation: some of then can predict secondary structure from your sequence, and then they look for PDB templates having similar secondary structure, even with low sequence similarity). In this situation, the template and your protein should be fold-related, and most surely they will still conserve spatially the most important residues (normally those from the active center). With this assumption, maybe you could still have a model of your protein. Try to use the different software proposed (I am most used to swissmodel).

Sure sir ill try it ....but can i publish a paper with that sequence.

I am doubt that by predicting structure of a protein using any software will make your work publishable. In case you are interested in understanding performance of various structure prediction methods visit http://www.predictioncenter.org/casp10/index.cgi

Hi dear

if u got a structure not elucidated with x-ray then you need to check out the structure homologous to your protein. You can do this in blast and align the sequences in clastalw. The overall homology of 2 protein is ok but you need to check their homology in the active sites especially coz after-all your aim is to model the active site, since sometimes we observe better over all homology but they fail in the active sites. Check the binding site amino acids of your protein in uniprot then check those particular amino acids while aligning the two sequences, i.e. Template and query protein. If the query protein binding site amino acids are identical or similar to the template protein then u can proceed for modeling the protein.\

you can download swift modeller and get the free license of modeller software. The full description for how to use swift modeller is mentioned on the site organized by swift modeller development team and you can proceed for your work........

As per my knowledge i have suggested you the way if any problem don't hesitate to contact me on [email protected].... I will be happy to help you...... Best of luck.....

I am fully agree with Dr. Raghava sir's advice. Since the sequence identity and query coverage both are poor, you should drop the idea of homology modelling and should try to predict the secondary structure of the target protein.

Dear as the experts have already mentioned that the sequence similarity is low so difficult to model the protein however u need to check out the similarity of the active site amino acids and if they are similar then u can model the protein because ulimately u need to model the binding site... Just try it

it need little trial exercise

Please try Itasser (http://zhanglab.ccmb.med.umich.edu/I-TASSER/) and then check your protein liability using online software and ramachandran plot.

Hope it will work to figure out the best predicted protein model.

In Conclusion i am decided to go for secondary structure of the target protein .

Thank you all

I have already seen excellent models (

Hi,

I agree with Benjamin Schwarz. Sequence Identity is not the only criteria to decide to model a protein or not.

First, define your objective/purpose why you want to model your protein of interest.

As you mentioned, Sequence Identity is near 30%, Did you consider the similarity and E-value with your template. Since the query coverage is also low, You cant model the full protein. I doubt that model wont be accurate for protein fold studies.

If the query coverage is in such a way that you could model a single domain of your protein. which may be relevant to the function of the protein, It may be useful for further studies, for instance, Drug Discovery projects. Let say its a functional/structural domain which may have implications it protein interaction region like Active Site/Ligand binding site.

In case of few proteins, modeling of Binding module domain would be useful to study the increased efficiency of the protein activity or to study its interacting partners.

I recommend what Dr.Raghava suggested. Please proceed with it.

And reg. publication, We need to understand the current research area of the protein of interest. Only homology modeling is not sufficient for publication in 2012 or in near future in so-called good journals.

If you are interested/planned to carry out experimental studies using the protein model, it would be really good and publishable too.

To summarize, If the model is helpful to provide insight on the protein, It is worth for publication. Please define your objective/need to model the protein then proceed.

All the Best.

Regards.