How to do phylogenetic analysis with 25000 sequences?

[Yevaud]

Hi,
I'm trying to employ 454 in phylogenetic research. We have 10 different sequences (markers) amplified among 19 species (ca. 1/16 plate) for the first try. I almost finished lab work and we are going to send the pooled samples soon. But I still do not know how I will analyze this data. I'm familiar with programs such as MrBayes or Paup, but the difference here is that in the end we will have many sequences and we expect to have hybrid species where some species will have more than one sequence. Therefore I guess I need two things:
- a platform to analyze sequences from 454 and exlude duplicate ones, get quality scores etc.
- a program to analyze that in phylogenetical context
Is anybody here aware about something that could be useful for me? Or maybe you can give me some tips?
By now there is one paper describing applicance of NGS in constructing phylogeny (Griffin et al., BMC Biology 9 (1), 19+) and they used galaxy, so I guess this will be my first try too. But on the other hand analysis they did with the data wasn't satisfactory.
I will be glad for any suggestions or advices! Thank you in advance!

[ajthomas]

You might want to start by simply separating the data by marker and species. Are you using MIDs and sequencing multiple samples together in each lane? You could use Roche's AVA software to separate everything out and combine duplicate reads into consensus reads. That will reduce the number of sequences you have to deal with and you may be able to use the programs you're familiar with.

*disclaimer* While I have done a lot of amplicon sequencing on the 454 and have used AVA quite a bit, I know very little about phylogenetic analysis, so my advice may be way off base.

[Palecomic]

QIIME should help you to do some aspects of what you want to do.

You can cluster sequences by identity using uclust (and other methods), pick representative sequences from the clusters and build distance matrices and trees from this. For amplicons other than 16S you would need to put together custom databases as reference for the aligners, but other people have been doing that for fungal sequences and the developers are active, friendly and helpful:

http://qiime.sourceforge.net/