For fulfilling the criteria of a good model, the target sequence should be atleast 40% identity or homology to the template sequence...

I think along with the sequence identity, the sequence coverage is also very much required.....

Sequence identity must be more than 30% for gud structure, otherwise you can go for fold predictions and then you can do the structure-sequence alignment..

Thanks all.

@Prasun: For my sequence coverage is good (~96%). I have predicted through I-tasser server also. It is taking the same pdb entry as a template. Any comments on that. Thanks in advance.

Hi Sourav,

The sequence identity is not a good score when you want to build a model of a domain. What is the homology score between your target and its template?

If it is higher than 35%, then it's possible to make a nice model for it.

I do not recommend I-TASSER. It has some problems in loops prediction and refinement.

Go for some offline softwares such as Discovery Studio, Schrodinger, MOE or ICM-Pro.

In my opinion a good model is a model that is useful [for designing experiments, informing hypotheses, rationalise sequence conservation etc], so it really depends on what you want to use your model for.

In practical terms I would use profile alignments and fold recognition to avoid or reduce template/target alignment errors. Then use template based modelling e.g. in modeller.

If more than one structure for this domain has been solved, look at structural variation. Should give you some idea where to expect variation [here your model in all likelihood will be least reliable] and which structural features are more conserved [you are probably closer to reality here].

@Milad I quite like I-TASSER based on its impressive CASP performances, but prefer Modeller when there is an obvious template for modelling [and importantly use my own alignment in modeller]. At 25% sequence identity I wouldn't worry about loop prediction [or bother much with loop refinement] as in all likelihood there will be inaccuracies in the 'core' already.

@Sourav

There are options in I-TASSER to exclude either certain PDBs (listed in a file) or all PDBs exceeding a sequence homology cutoff (for example, all templates >60% homologous)

Dear Sourav,

For a good homology modeling the sequence identity should be always above 35% otherwise there will be some changes in the protein beta sheets or alpha helix. Mostly it depends on the homology between the template and your sequence. If the identity is below 30% then there is nothing to worry if you found good template having query coverage area above 90% then there are better chances of getting good model. Just you have to refine your model (loop refinement) for that you can use Modeller, I-Tasser, Discovery studio.

I agree, sequence identity should be over 35% for a relatively good model. Below 30% is considered to be in the "twilight zone" and most methods have significant difficulty predicting below that threshold. I think using I-TASSER is a good idea though for the prediction of this model and will likely give you the best guess, but it will likely be ~20-70% accurate.

Dear Sourav, i will suggest you to do all combination i.e Ab initio, threading and homology modeling. If you get a similar stable fold or domain in all cases then you can atleast do preliminary structural study.

Feasible homology cutoff is 30%

Article Advances in Homology Protein Structure Modeling