Don't know if I'm right

[anna_]

Hi everyone,

we have just sequenced a whole trancriptome of a conifer de novo and now it is my challenge to find candidate loci in there. I want to perform a stand alone blast search. I'm so new to the subject that I don't even have a concrete question.

Anyway - hello to all!

[westerman]

Well, if you are new to sequencing then I suspect that you do not have access to a stand-alone blast server. While it is fairly easy to set one up, you may wish to just try Blasting several of your transcriptomes to a public blast server (e.g., America's http://blast.ncbi.nlm.nih.gov/ or Europe's EMBL http://dove.embl-heidelberg.de/Blast2/ ... and others)

Actually for a one-off project I would suggest just using a public resource instead of your own server unless you have special requirements (e.g., Blasting against a unusual database). The public resource may be slower than your own server but it will not be that much slower and will be much easier to use.

[anna_]

Thanks for that quick comment.
Our transcriptome has 500 000bp, the LargeContig FASTA-File has 7.5MB. I want to find protein coding sequences in there. The plan is to start with proteins from Arabidopsis and Rice. Therefore I think I need tblastn. I guess that can't be done online. I had the idea to format my FASTA-file to a database and blast the know proteins of the plants against it.

At the moment I'm trying to install the executables available at the NCBI site.

[anna_]

Sorry, I meant 500 000 reads (454)

[westerman]

tblastn can be run on-line as well as blastx. To do what I suggested you would use blastx -- search the rice and arabidopsis databases with your nucleotide 454 sequences, translated. Your idea of using tblastn to look through your translated 454 sequences with the rice/arabidopsis proteins is also good.

With 500 000 reads it is probably is better to install your own local blast. As I said it is easy enough to do, at least from the command line. Spot checking your results with the results you can get from a public site is recommended. Depending on the power of your computer the blasting can take many hours or days. I suggest reducing your data set to 5000 or so reads for testing purposes before you try all 500 000.

You may also wish to consider a program that can take splicing into account (e.g., splign).


If you do come up with a concrete question then post on SeqAnswers. There are many knowledgeable people here.