search genes in a reads pool or (very) draft assembly: any usefull tool?

[Gorbenzer]

Hi everyone,
i'm in the need to screen a few bacterial genomes for the presence of around 100 genes i defined.

We have not much sequencing data for this genomes (20MBp circa with 200bp reads), that is far from the quantity needed for a good assembly.

I've already tryed to blastn/tblastx my genes database against reads and the contigs coming from the very draft assemblys we can get with this data but i'm not satisfied with the results because of the too high ration of false positives.

There's a tool that can help me in this work??? maybe some kind of metagenomic tool?


Thanks in advance to anyone that will help me

[gaffa]

I haven't tried it myself, but maybe this could be of interest: http://www.plosone.org/article/info%...l.pone.0019816

[krobison]

TBLASTX is a good way to get a lot of difficult-to-interpret noise; I strongly recommend using BLASTX against a carefully focused database of proteins. If you are concerned with false positives, tighten up your P-value cutoff.

Merging your reads with FLASH prior to assembly may boost things a bit, if you have paired-end data. Still, depending on your bug, 20Mb of reads should be 2-10X coverage and you would be expected to get some sizable contigs. What sort of contig size distribution are you getting?

Also, it is relevant which sequencing technology you are using. Ion's indel issue could be causing you much grief, if that is the technology.

In the end, what is the question you wish the data to answer? It may be that this dataset is simply too sparse. On the other hand, for some questions it may well be good enough.