Verify protein sequence

Question

[bzhtgb]

Hi all,

I have to verify a protein sequence from 2 insect species. The genome is not very clean for one of them. For the second species, the genome is not available.

What is the standard process to achieve that?

I think I have to :

• compare my protein with orthologs from others species, Length of the protein,... That could provide a warning: false CDS? ORF prediction in the genome. For the second species, no genome...

• others idea are welcome!

Best regards

Answer 1

[Michael]

I think RACE-PCR and getting a full length consensus cDNA is the best approach to date to get reliable gene models. This works with or without a reference genome. Alternatively, de-novo peptide sequencing but I think this is more complicated.

If validation means only having similar structure and an antibody is available, then western blot should do it.

[bzhtgb]

Thanks Michael, but I can only use in-silico method at this time.

[Michael]

Ok, but IMHO it is not possible to get something close to counting as 'verified' from using in-silico methods only. You can try all homology based methods (blast, exonerate), you can make a phylogeny and see if the proteins fit, you can check the presence of the correct domain structure using InterProScan, and you can do 3D, 4D reconstruction using e.g. SwissModell or i-Tasser checking the results of TM-align. See e.g. this paper by Daly et al. This is as close as it gets, and still you will never get a validated sequence, because you could be missing that one little exon only PCR can give you.

I recommend to find a collaborator to do the experimental lab work with you, these are really routine tasks which give you good return on investment, and any stakeholder in your project should have a high interest on getting real validated sequences.

[bzhtgb]

Thanks Michael, Very useful information. I agree with you. I will perform an alignment with closest species (drosophila) only to have an idea of difference in sequence length, mutation, ...That will give me some arguments to suggest a PCR.

best

[Michael]

Btw, if you let us know which genes we are talking about, it might be possible to help you even more.