SWISS-model, i-TASSER are most widely used and reliable online servers for protein modeling.

Dear Yedukondalu Kollati ,

If I understand you correctly you mean that the 3D structure(s) of the mutant(s) is/are not available since the WT is (https://www.uniprot.org/uniprot/P01266 ).

Indeed, as said by Tushar Chakraborty you can use I-Tasser for the best possible prediction: https://zhanglab.ccmb.med.umich.edu/I-TASSER/

However, I am afraid that the overall structure (most of the times) will be quite similar. However, looking at for example:

Miller, D. B., & Piccolo, S. R. (2020). Compound Heterozygous Variants in Pediatric Cancers: A Systematic Review. Frontiers in Genetics, 11.

https://www.frontiersin.org/articles/10.3389/fgene.2020.00493/full

It states “Common variant prediction algorithms, such as SIFT and PolyPhen-2, use nucleotide sequence homology (among other parameters) to predict whether protein function is affected by an amino acid substitution (Ng and Henikoff, 2003; Adzhubei et al., 2010). Predictive algorithms often have low specificity for missense variant prediction (Richards et al., 2015).”.

In for example (see also enclosed file):

Liu, S., Zhang, S., Li, W., Zhang, A., Qi, F., Zheng, G., ... & Ma, X. (2012). Clinical and genetic analysis of a compound heterozygous mutation in the thyroglobulin gene in a Chinese twin family with congenital goiter and hypothyroidism. Twin Research and Human Genetics, 15(1), 126-132.

This paper has a (small) bioinformatics section as well. These papers might give you some ideas for further analysis.

Best regards.

Dear Rob sir,

Thank you for sharing valuable information.

I have found one Compound Heterozygous Variants in TG gene. I have predicted the variants in SIFT, Polyphen-2 and PROVEAN to find the deleterious. So, I would like to do 3D modelling. The TG protein contains 2768 amino acids. I want to do modelling from 553 amino acid to 2052 amino acid (total: 1500 amino acids).

If I use I-TASSER or SWISS-model, shall I give first wild type and then Mutant protein sequence?

or

Shall we show compound heterozygous variants at a time in 3D modelling?

These tools are enough to do protein modelling.

Thank you very much sir.

Dear Yedukondalu Kollati ,

Ah I see, well you can have a go with I-Tasser. The program will predict the 3D structures of your mutants. In principle the WT info is already known so no real need to ‘predict’ WT first (but in order to see if the program predicts something else when you take a ‘shorter’ version of the protein might be interesting).

It will be interesting to see if (and to what extend) the 3D structure changed due to the mutation(s). I suppose you have to take it from there.

Good luck and best regards.

PS. In order to use I-Tasser you need to register (no worry: no cost behind it) and be aware it can take several days before you get the results in your mail box (the server is quite popular). I suppose you are familiar with for example Chimera in order to visualize the pdb files you will get from I-Tasser. See for a bit more details on this:

Article The role and significance of potential lipid-binding regions...

Data keller3biotech2015suppl.docx

Before starting any homology modelling, first check the templates in the Blastp (protein blast) to construct the model for your query sequence using the pdb as the search set. If you get more than 40 percent sequence similarity, then you can continue with homology modeling. If you have Schrodinger, then you can use this suite's PRIME homology modeling module.

I-TASSER, SWISS-MODEL, or Phyre2 servers can be used otherwise. Modeller can also be used. It is a standalone program.

If sequence similarity is very less then you need to proceed with threading-based or ab-initio modeling of your protein.

One can also perform the secondary structure prediction prior to modeling. This can be done from PSIPRED or JPRED, etc.

There are different types of modeling, single template, or multiple template-based modeling. So, prior to modeling, one must research well on its query and templates and which part of the protein you need to model. Sometimes it is not necessary to model the whole protein if the sequence similarity is very less. Only a required domain can also be modeled to start the study with.

The quality of the protein structure should also be verified by the Ramachandran plot or from the PROCHECK server.

After building the model, refine the loops, and minimize the structure.

Molecular dynamics simulation can be performed to equilibrate the model and obtain a stable and relatively low energy state of the model.