We need a coherent NR for KBase. Not one for microbes and one for communities. Sent from my iPad On Jan 10, 2013, at 3:24 PM, Folker Meyer <[email protected]> wrote:

Dylan,

good idea. actually this is a great service for people wanting to search metagenomes, I encourage you to build it.

It should do exactly what you suggest:

- any sequence input is searched against the m5nr - retrieve md5s for the m5nr entries - retrieve list of metagenome_IDs that this md5 is present in - find any sequences matching the md5s in your metagenome(s) of choice with very loose cut-offs - re-run a BLAST(?) or some other search on the sequence set and return things that hit for real with good (strict) cut-offs.

We will have to implement an Md5->metagenome_ID function (we can do that if you decide to go for this)

I think this would be an excellent service to offer, it might be resource constraining, but this is definitely something that a lot of users have asked for in the past.

Best, Folker

On Jan 10, 2013, at 12:07 PM, Dylan Chivian <[email protected]> wrote:

...

... and what about searching metagenomic sequence? (i'd wink, but it's not funny)

since these guys use the m5nr, we could expand any hit from a query sequence to an m5nr sequence into the mg reads that are close to the m5nr exemplar. just a thought. any other ideas?

thanks, - dylan

On Wed, Jan 9, 2013 at 8:55 PM, Christopher Henry <[email protected]> wrote: Well, it's a fair question.

First, we are going to have users, like me, who specifically want to run specific BLAST queries for specific reasons. For that, I think we need a BLAST server. I for one want a blast query to look for specific sequence without having to worry about whether or not the gene callers missed this gene. In particular, this is something we are speculating is a universal gene, and so for the genomes where we don't find it already, we want to do a REALLY thorough search. Does this make sense? So I was thinking TBlastN is what I want… even if it is painfully slow and inefficient… it's to meet this very specific need… and I won't be using it to process a ton of sequence. Just query 10 or so diverse versions for this protein.

I posit… given the increasing drive towards protein speculation and resolving more challenging problems in annotation, this is going to be a common use scenario.

Second, it's completely fair to say we need a service that says "I have a sequence that I want to find what KB knows about". It's a separate use scenario in my mind. It's almost a search query, and the sim service is darn close to being able to handle this particular use scenario.

On Jan 9, 2013, at 2:43 PM, Robert Olson <[email protected]> wrote:

...

Do we really mean "BLAST" in particular, or is the problem really "I have sequence that I want to find what KB knows about" ?

On Jan 9, 2013, at 2:41 PM, Shreyas Cholia wrote:

...

OK - that was some serious redundancy, but I think you catch my drift.

application specific syntactic sugar for specific apps.

_______________________________________________ Kbase-devel mailing list [email protected] https://lists.kbase.us/mailman/listinfo/kbase-devel

_______________________________________________ Kbase-devel mailing list [email protected] https://lists.kbase.us/mailman/listinfo/kbase-devel

_______________________________________________ Kbase-devel mailing list [email protected] https://lists.kbase.us/mailman/listinfo/kbase-devel

_______________________________________________ Kbase-microcomm mailing list [email protected] https://lists.kbase.us/mailman/listinfo/kbase-microcomm