Find genes in contigs based on homology proteins

[thedamian]

Hello,
I have contigs from a new organism (its genome isn't known well).
I know, that this organism produces some proteins that are similar to proteins of other organisms.

I would like to find an approximate location of a gene that produces a protein similar to other homology protein.

Do you know some good tools or pipelines for that purpose?
A have tried tblastn and SWiPS however I am little confused with the result and don't know how to interpert it
(how to get a linear sequence of supposed gene)

Does anybody have some good advices or an experience in this area?

[krobison]

How many contigs? How big are they? How were they generated?

BLASTX is a straightforward way to approach this. Make sure you set a low limit on the number of overlapping hits; otherwise a lot of hits in one region will prevent hits from being reported in another region.

If you have very long contigs & think that frameshifts are rare, you should use a gene finder apropos to the domain of life you are in (for example, Glimmer3 is good for eubacteria & archea but doesn't model introns).

You should start by reading some genome sequencing papers of similar organisms; this will give you an idea of the sorts of tools that might be applied.

[thedamian]

Originally Posted by krobison

How many contigs? How big are they? How were they generated?

BLASTX is a straightforward way to approach this. Make sure you set a low limit on the number of overlapping hits; otherwise a lot of hits in one region will prevent hits from being reported in another region.

If you have very long contigs & think that frameshifts are rare, you should use a gene finder apropos to the domain of life you are in (for example, Glimmer3 is good for eubacteria & archea but doesn't model introns).

You should start by reading some genome sequencing papers of similar organisms; this will give you an idea of the sorts of tools that might be applied.

I have 27050 contigs that have length of 1000 or more bases. The longest is 90691. So lenghts are good. They were generated by De Novo assembler embedded in CLC Genomic Workbench (reads are from MiSeq).
My organism is roundworm so introns exist.

I know some basic theory, but I'm struggling with applying it.

[A_Morozov]

If you only need several genes, you can gather them by hand. Download some sequence browser (I personally prefer artemis), look up hit location, strand and frame in blastx/tblasn output and copypaste aminoacid sequence. Of course, if gene contains introns you will get some uncertainty near exon boundaries, but at least you will have most of the sequence and hopefully active sites.
If, on the other hand, you want to find ALL genes in a genome, considering all splicing and detecting pseudogenes and whatnot, I advise you to google JGI and TIRG gene finding/annotation pipelines. But that will cost a significant amount of computational power and you will probably want to outsource this part of work to bioinformatician if you don't have one in your lab.

[thedamian]

I want olny five genes, so this simple approach seems to be fine. I'll check it