Pulling out homologous sequences from different transcriptomes

[mruhsam]

Hello,

I have the assembled transcriptomes from 15 different (more or less closely related) species. I would like to pull out all the (presumeably homologous) contigs which occur in all the species so I can align them and build a phylogeny. I blasted one species against all the other ones and got a 5GB file but I don't know how I can pull out the contigs of species 1 which have a hit in all the 14 other species....

Is there a program that can do that and give me a file with all the contigs which have a hit in all my species?

Thanks for your help

Markus

[sarvidsson]

Sorry, overlooked that you were talking about transcriptomes. For transcontigs, standalone InParanoid is worth a try - limited to protein coding transcontigs, however.

[mruhsam]

Mauve is a good program to do the alignment but at the moment I don't have the input file to use Mauve, i. e. I don't have a file with all the contigs which occur in each of the 15 species. I don't know how I can pull out those contigs to create an input file in the first place.

[sarvidsson]

See the edited post

[mruhsam]

I actually had a look at InParanoid but got the impression that my data are compared to the data in their database. Or did I miss something there, i. e. can I use InParanoid to create pairwise comparisons between my data sets?

[sarvidsson]

I actually had a look at InParanoid but got the impression that my data are compared to the data in their database. Or did I miss something there, i. e. can I use InParanoid to create pairwise comparisons between my data sets?

You can make your own pairwise comparisons - but you need the standalone version (bottom of the page).

[GenoMax]

@mruhsam: Can you give us an idea of some numbers? How many contigs per species (or are you referring to transcript models assembled as "contigs", so thousands)? What is the size range on those contigs?

[mruhsam]

There are well over 100.000 contigs for each species varying in length from 200 bp to about 20.000 bp.

[GenoMax]

You may want to start with the largest of the lot (assuming they were assembled reasonably correctly) say 10K bp or more. If the species are very similar then you could use stringent search criteria. It sounds like depending on what you find you are going to need to some custom parsing of the results.

If you are able to convert these into protein sequences then OrthoMCL (like InParanoid mentioned above) would come in useful: http://www.orthomcl.org/orthomcl/