Dindel problem: overlapping windows and non-uniquely re-aligned reads

[Yilong Li]

Hi all!

We are trying to use Dindel for generating re-aligned .bams from Illumina PE-read data. I was able to run through the Dindel pipeline, but noticed that in the output .sam files (generated following the instructions given in Dindel manual), many reads were output more than once.

When examining my data, I realized that overlapping windows were created in Dindel Stage 2. (The makeWindows.py step.) This is probably the reason why I was getting multiple reads in the concatenated .sam files.

Has anyone else encountered the same problems while attempting to get re-aligned .bam files? Do you know any solutions? Merging windows would be an obvious option, but could this result in large windows with unmanageable number of distinct haplotypes?

Any comments are welcome! 8)
Yilong

[csoong]

just curious, are you outputing SAM to look at the indel sequences? I never became aware of that function. Thanks.

[gaffa]

There is a warning about this in the Dindel manual (though no solution).

[Yilong Li]

Thanks for the replies csoong and gaffa.

I want to output the re-aligned reads in a new .bam file to visualize them in a in-house program. Also we could use the re-aligned .bams in GATK UnifiedGenotyper. The actual indel sequences can be seen in the VCF file produced in the last step of the Dindel pipeline I think.

gaffa: yeah I've seen the warning in the manual. I am wondering if anybody been able to combine the re-aligned reads into a single .sam file.

[Yilong Li]

Bump. Please share any ideas!

[libiyagirl]

Hey I merged all the windows from dindel into a single bam file using samtools merge command. But I didn't check if there's any overlapping windows. Do you always have this problem?