I have the an assembled genome (contigs) of an insect that is not yet annotated. There are some genes of interest (such as actin) that I want to know if they are present in that genome. What can I do
•
written
by
christian_jpg2 •
Thanks.
However, I already try that with a sequence of the closest relative that I can find (they are from the same subfamily), and I do not get any hit. I do not know if their sequences are that different or if I am doing something wrong
what kind of sequences are you using as input? nucleotide? protein? and related: what kind of blast are you running?
christian_jpg2 : As has been suggested in this thread you could try tblastn if plain blastn did not work.
You also have to consider the possibility that if you did the blast search right and did not find a hit for something that should be present then your assembly could be of poor quality. You could take your assembly and try blasting it against nr to see if you get reasonable/contiguous hits.
It looks to me, as if you should follow this thread "Building Hidden Markov Model (HMM) for proteins" closely, as both of you want similar things.
As for HMMs of your proteins, you could retrieve them using PFAM. The actin page, for example, is here: https://pfam.xfam.org/family/PF00022
Then, you can follow this EBI tutorial to retrieve the HMM from PFAM.
Actually, what I would do is very crude and a bit dirty: I would translate my genomes in all frames and check with the downloaded HMM using HMMER, if I get a hit. It would be great, if HMMER could use protein HMMs to search DNA, but this functionality does not yet exist (I think).
running a simple tblastn with protein against the genome will be more then sufficient here.
For more specific (or really distantly related species == very low sequence conservation), your given approach will indeed be the appropriate one.
Yes, lieven.sterck is right, obviously. I had assumed, genomax already did that. Upon re-reading, I saw, that probably BLASTn has been used, so I agree, tBLASTn would be the keyword.
Log in to answer this question.
