Finding if Sequence is Present in Long List of Other Sequences from Multiple Species
Hello,
I have a sequence of 500nt, that has an important function in Species X. I have a list of about 20 other species, with coordinates of about ~20,000, that I want to see if they contain this important 500nt region. I did an initial BLASTN between this sequence and the 20 other sequences, and I haven't seen anything significant. I went and did a BLAST for "Somewhat similar sequences", and I'm getting hits, but they're very short, and some have not the greatest E values.
What are some ways to make my search more thorough so I can more confidently state that these 20 species do not contain this important 500nt sequence.
0 answers
No answers yet.
Log in to answer this question.
More posts like this
-
Has this sequence really more then 70k perfect matches in the human genome?
written by gernophil •
-
BLAST parameters for sequences with high similarity
written by behzad.karkaria •
-
Including "no significant hits" in output of local blastn searches
written by newol210 •
-
Puzzling NCBI BLAST+ output after transcriptome assembly
written by ella •
-
How to get the sequence differences between multiple bacterial genomes
written by Rose •
-
Blast command line gives no hits
written by Vca80553 •
-
How can I know that the blast hits of my NGS data mapping against my query sequence?
written by apl00028
-
too many transcripts after trinity
written by mxlsherry1992 •
-
How to set up a reference file for BWA so that highly variable sites between conserved regions are …
written by DNAngel
-
comparing two genomes
written by dr_mm_sh •
You have to consider the possibility that unless this sequence is regulatory in nature, it may not be under strong selection pressure. If it is protein coding you may want to do protein level comparisons since those are always preferable to nucleotide.
Thank you for the reply. Unfortunately, we the sequence of interest is an enhancer region, so there are no protein level comparisons available.
related to genomax remark: How important are each of the 500nt ? I mean, if they all 500 have to be there then doing the blastN as you did is a valid (best?) and accepted way to prove that the sequence is not present in the other species.
What do you mean with "somewhat similar sequences"?
It's hard to say; since the 500nt sequence is from a different species, so I don't expect all of it to be intact. I was expecting more than just small (<20 nts) matches however.
"Somewhat similar sequences" is a setting you can select when you are doing a BLAST.
ah, had to look it up and this "Somewhat similar sequences" is just an option to run the 'normal' blastn algorithm ... I must be getting old :/
If in those of 500nt of promoter(?) region you find stretches of up to 20nt of exact(?) matches then you can start to conclude there might be some conservation between those sequences.
This is also depending on the evolutionary distance between the species used in your analysis: if they are closely related you might find those matches randomly (== no biological conservation) as well though.
If it’s an enhancer region, the specific sequence is likely less important than its secondary structure or sequence pattern properties (repetitiveness but without any specific base sequence for example).
You may want to consider searching with HMM based methods instead.
Thank you jrj. I'm not extremely familiar with hidden markov model search methods. Do you have any additional resources I could look at?
I would just start by taking a look at the HMMER website. They’re a little more popular in the protein world, but they’re still useful for detecting remote and structure-based homologies.
I think the site has an FAQ. Otherwise, any one of a number of reviews by Sean Eddy.
I was thinking along the same lines but I believe it would require at least a couple of sequence to be able to build a sensible HMM profile?
If there is only a single sequence available anything other than blast will become less feasible I'm afraid.
Yeah I was sort of implicitly hoping OP might be able to curate a couple of representatives manually to get started perhaps.