BreakDancer output question

[cwhelan]

I've been using BreakDancer for structural variation, and it works pretty well, but I'm a little bit confused by some of the output. My understanding is that BreadDancer categorizes translocations as "CTX", but sometimes in my output I get lines like this:

1 121473724 521+551- 7 61995061 374+334- DEL 1822 99 230 LC_1|230 -38.93

Has anyone out there seen anything like that before? I was expecting that any calls that had coordinates on different chromosomes would be of type "CTX", and I'm not sure what a deletion ("DEL") with breakpoints on two different chromosomes means.

Thanks,
Chris

[kalinka23]

Hi!

Sorry that i don't have an answer but a question instead... i am very new in using breakDancer and i don't know how to put in the order of normal and tumor bam files if i want to detect structural variations in a tumor sample compared to the matching normal sample (or isn't it possible?)

According to the pipeline i did the following:

perl ~/bin/breakdancer-1.1_2011_02_21/perl/bam2cfg.pl -g -h tumor.bam
normal.bam > P1_BreakD.cfg # (i used as tumor.bam:
s_110825_4.bwa.hg18.bam.sorted.bam and as normal.bam:
map:s_110825_3.bwa.hg18.bam.sorted.bam)

in the P1_BreakD.cfg i see the following
readgroup:NA platform:illumina map:s_110825_4.bwa.hg18.bam.sorted.bam readlen:101.00 lib:NA num:10000 lower:148.66 upper:268.33 mean:206.65 std:14.95 SWnormality:-13.76 flag:0 exe:samtools
view #THIS IS MY TUMOR SAMPLE!!!
readgroup:NA platform:illumina map:s_110825_3.bwa.hg18.bam.sorted.bam readlen:101.00 lib:NA num:10000 lower:157.07 upper:279.52 mean:215.25 std:15.28 SWnormality:-14.92 flag:0 exe:samtools
view #THIS IS MY NORMAL SAMPLE!!!


i used this file (P1_BreakD.cfg) to calculate the variants in the
following way:
~/breakdancer-1.1_2011_02_21/cpp/breakdancer_max my_BreakD.cfg >
P1_BreakD.ctx &


in the output file (my_BreakD.ctx) i see e.g. the following:

head P1_BreakD.ctx
#Software: BreakDancerMax-1.1r112
#Command:
/home/schlange/bin/breakdancer-1.1_2011_02_21/cpp/breakdancer_max
P1_BreakD.cfg
#Library Statistics:
#s_110825_3.bwa.hg18.bam.sorted.bam mean:215.25 std:15.28 uppercutoff:279.52 lowercutoff:157.07 readlen:101 library:NA reflen:3017206663 seqcov:12.1111 phycov:12.9055 1:1470 2:27634 3:158709 4:33019 8:1278 32:43684
#Chr1 Pos1 Orientation1 Chr2 Pos2 Orientation2 Type Size Score num_Reads num_Reads_lib s_110825_3.bwa.hg18.bam.sorted.bam s_110825_4.bwa.hg18.bam.sorted.bam
chr1 921503 8+8- chr1 921568 8+8- INS -75 99 8 s_110825_3.bwa.hg18.bam.sorted.bam|8 NA NA
chr1 954920 3+0- chr1 955267 0+5- DEL 379 99 3 s_110825_3.bwa.hg18.bam.sorted.bam|3 0.11 NA
chr1 1638956 2+0- chr1 1639003 0+2- DEL 74 61 2 s_110825_3.bwa.hg18.bam.sorted.bam|2 1.05 NA
chr1 1856102 4+2- chr1 1856125 4+2- INS -78 61 2 s_110825_3.bwa.hg18.bam.sorted.bam|2 NA NA


and i find only lines with the name of the normal bam, and also in the
headers only the name of my normal.bam appears
(s_110825_3.bwa.hg18.bam.sorted.bam). But i wanted to determine the
variants in the tumor sample compared to the normal sample. Is this the
case, do i have in the output "P1_BreakD.ctx" the variants in the tumor
sample based on the comparison to the normal sample? Or how can i
understand this, and if this is not the truth: what do i have to do in
order to determine the variants in my TUMOR sample?

And, regarding the explanation of the output columns, where can i find the
most comprehensive documentation of that?

Thanks a lot for any help!!!

Best,
Karin

[CATCH]

Hi Karin,

Did you find an answer to your problem?
I find the documentation a bit confusing, because it does look like the example specifies tumor and normal bams for detection (http://gmt.genome.wustl.edu/breakdan...mentation.html) but trying the same and looking at the output, it seems to me that these are the common, i.e. the germline SVs, that are reported this way.

My next attempt will be to run BreakDancer for the tumor and normal separately and overlap the resulting breakpoint positions (within some hundred basepairs) to find breakpoints that are unique to the tumor.

-Fiona-

[Irsan_Kooi]

Hi there Karin and Catch,

I have the same questions as you. Right now, I am pursuing the same strategy as karin to find out the answer (run breakdancer on tumor and normal sample seperately and look at concordant and discordant SV's). Karin, have you finished comparing the two seperate breakdancer runs and compared them?

Cheers,

Irsan

[Irsan_Kooi]

Hey all,

I found out that when you want to do a tumor normal SV-detection with breakdancer you need to be sure that your have the read group information in your bam-file set. To check if that is the case do:

samtools view -H /path/to/my.bam | grep '^@RG'

If there is no output, there is no read group information. You can add it by the picard-tools AddOrReplaceReadGroups like this (make sure your .bam is sorted!):

java -jar /path/to/picard-tools/AddOrReplaceReadGroups \
INPUT=your_sorted_bam_file.bam \
OUTPUT=your_file_with_headers.bam \
RGID = 'your read group ID' \
RGLB = 'your read group label' \
RGPL = 'your read group platform' \
RGPU = 'your read group platform unit' \
RGSM = 'your sample name' \
RGCN = 'the name of the sequencing center' \
RGDS = 'some description for you project' \
VALIDATION_STRINGENCY=LENIENT

[vinay052003]

Also, just to bring it to the attention, each read group has to be unique for each library.
If you have same readgroup id (RGID) for two libraries, breakdancer will not be able to perform correctly.