problem aligning SOLiD reads with BWA

[rhys]

Hi All,

I recently got some 50nt SOLiD reads. The 53X deep dataset I have been focusing on is from exactly the same isolate as a reference genome. However, only 9% of the reads are aligning! and 1/3rd of the genome is uncovered. By trimming the reads I can get up to around 50% of the reads to align. Apart from this, which might be a problem with the sequencing?, my pileup's look unusual in the fact that all the reverse reads, instead of showing up as commas if they match, are showing up as a lower case letters, I have attached a picture of the tview showing it.

I first changed all -1's to 0's in the quality file, then indexed, ran BWA's solid2fastq.pl and aligned.
A messy way around this is to compliment all lower case letters, but I was wondering if this is meant to happen?, and if anyone knows of something that I'm doing wrong?
I haven't worked with SOLiD before so any advice would be useful.

Thanks,
Rhys

[Chipper]

Hi,

which version of bwa do you use and what number of mismathes? Try something like -l 25 -n 8 if you have low quality 3' ends.

Other options that might give you better results would be bowtie (faster, even with 3 mismatches in the seed), BFAST (especially if you have errors or N:s in the 5' end) and perM.

[rhys]

hi chipper,

Thanks for the reply
i'm using v0.5.5.
I might try another alignment program if changing the settings doesn't improve the alignment.
however if I can get the pileup to show compliment reads as commas, then I can parse it a lot more easily. Do you know if the BWA pileup is meant to look the way it does in the attached picture?

[Chipper]

bwa 0.5.5 does not complement the reverse reads for SOLiD alignment, you can fix it with a short perl script but since this bug was fixed in later versions it might be easier just to install the latest release and realign it. Your tview picture shows that all reverse reads are reversed but not complemented, so tview thinks it is mismatches thus shows the sequence rather than commas.

[rhys]

Great! so I've downloaded the newer one, and will realign the reads. Then I've just got to work out why so few of the reads are aligning. I will post back if I find anything interesting.
Thanks,
Rhys