Gene duplication, as we know it, is because of exact copying of the genes, in a genome... If you have the nucleotide sequences of the members, perform blastn for pairwise, or perform multiple sequence alignment, clustal, and check, how close they are. If the identity is too high, this is suggestive of gene duplication. I hope this helps.

Dear Anubhav,

Thank you very much for your suggesting and are there any criteria for this, for example, how many percentages it should be identical?

Hello Abdugaffar,

It's better, if you optimize your own protocols. Take nucleotide sequences of known proteins and families, which are known to show phenomenon of Gene duplication, and perform the training experiments... Decide the cut-offs based on that. Try to have a large dataset, to convince the reviewers. Consider subject coverage and query coverage also. In order to get best hit, consider, Bit score. Once, optimized, use the cut off on your dataset.

Dear Anubhav

Thank you again for your consideration

Hi Leopoldo,

I think, tblastx is not a good option as in tblastx, both, the query and subject are first converted to protein and then aligned, which is not required here, and is a source of false positives. for gene duplication, identity in nucleotides is to be seen. 

I think this tool will be very helpful for your problem http://www.ensembl.org/index.html

Thank you for your suggestions