how do i find a specific sequence in a bam file?

[volavii]

I want to extract special sequences out of my bam-file (reference-mapped with BWA).

normally i do that with blast or blat, but this time i have a bam file, not a ready-to-use genome...
do i have to assemble the mapped reads into a consensus sequence in bevore, or is it possible to first (1.) identify the respective scaffold via the reference-genome with blat, (2.) assemble the reads that mapped to this scaffold and then (3.) aligne my sequence to that assembled scaffold?Or is this idea totally stupid? :/

I have never done an assembly so far. Iam really unsure what is the right way here...

Which tool would you suggest for assembling a bam-file, when dealing with genomes of >2 GB? And how shoud I care fore heterozygous positions?


so many thanks in advance,
hope anyone can give me here some help

[westerman]

It is hard for me to understand what you need. It appears that you already have reference-mapped data thus denovo assembly does not seem to required. If instead you are asking either (a) how to extract reads of a certain region or (b) how to call SNPs/Indels then you should look at the samtools 'mpileup' command. See http://samtools.sourceforge.net/mpileup.shtml for a starting place on mpileup.

[amarth]

Have you tried to convert your BAM file into a SAM or Fasta?

Use in the command line:

$ samtools view -h -o out.sam in.bam

The SAM file will provide the mapped reads into a specific scaffold. There you can retrieve your reads. Then you can assemble the reads

[volavii]

first of all, thanks for answering!

sorry for my imprecise question.

What i have done so far is a reference-mapping of 5 genomes. i used BWA for that. everything worked well.

Now i want to have the sequence of ~25 genes outof these 5 inidividuals. And i am not really sure how to do it.
I have no experience how much time it takes to assemble the reads of these 5 individuals to a consesus sequence... therefore i thougth: maybe it is enought to just assemble the parts of the individuals in which the genes are lying...

@amarth: which program do you suggest for the assembly. And how should i care for heterozygous sites? Shoud I use ambiguity code in the final gene sequence, or "Ns"?

Thanks!