sequenceserver
sequenceserver copied to clipboard
wurmlab
slider with 3 pre-determined "search stringencies"
Provide a slider that lets users choose between 3 presets. These would populate the advanced params:
- normal --> default
- highly similar sequences (specificity?) --> Evalue cutoff at e-3; BLOSUM80, longer wordsize?
- distantly related sequences --> BLOSUM45 matrix?, shorter wordsize
(other params may be appropriate too).
We need one setting for "short highly similar sequences" (primers for DNA query or short peptides). It came up here: https://plus.google.com/u/0/106914048811184527917/posts/aiZX47dewcw
And NCBI do give valid recommendations here: http://www.ncbi.nlm.nih.gov/blast/Why.shtml
We need one setting for "short highly similar sequences" (primers for DNA query or short peptides). It came up here: >https://plus.google.com/u/0/106914048811184527917/posts/aiZX47dewcw
This URL doesn't resolve for me..
And NCBI do give valid recommendations here: http://www.ncbi.nlm.nih.gov/blast/Why.shtml
Sounds good to me.
probably one of those circle issues. - maybe you need to be in the circle of the guy who created the post.
In any case, someone wanted to check where his primers mapped in the genome. BLAST sucks for that by default….
I often blast (non-degenerate) primers against transcript databases and it works fine, actually. Don't get great e-values but it gets there with a bit of manual inspection. Still, probably could be better, so I agree, particularly since NCBI gives specific settings that can be used for the purpose and we don't have to experiment.
Whoa - just realized that NCBI changed the default on us yet again.
-task <String, Permissible values: 'blastn' 'blastn-short' 'dc-megablast'
'megablast' 'vecscreen' >
Task to execute
Default =`megablast'
Nice thing: they have a built-in '-task blastn-short' which is probably appopriate for primers. One bad thing is that default blastn is now megablast (which is insensitive beacuse it implies min_wordsize 25 (or so)). This is misleading - I always thought blastn was blastn.
Nicely noticed. Should we be opinionated the default settings to be blastn instead? I don't see why not.
Yes.
Especially because people are more likely to use seqserv for non-model organisms (and thus more distant sequence similarities) than what megablast was intended for)
Not sure if you understand me - I was trying to say that blastn is rarely used anyway for searching when query and db are from distantly related organisms. Of course there is exceptions to this, but maybe it doesn't matter that much, because if you are comparing nucleotide to nucleotide then by definition you probably aren't using disparate species. But anyway, I still agree we should default to blastn because that's what we say we do, so feel free to ignore me.
Sorry.
I think it sometimes is used. Typically when people are trying to design primers across species. (e.g.: for a species that is between two sequences species).
@yannickwurm writes:
*** General search options -task <String, Permissible values: 'blastn' 'blastn-short' 'dc-megablast' 'megablast' 'vecscreen' > Task to execute Default = `megablast'This is misleading - I always thought blastn was blastn.
Fixed in next (SHA 4fae0c839e5145046060de58d59eec2675930b4c).
