breakdancer output question

[csoong]

Dear Readers,

I pasted 2 lines of breakdancer output below. According to the manual column 2 and 4 are for recording the reads orientation information that mapped to the anchoring regions. But form the 2 record lines, which are in close vincinity, why is the orientation info different between the 2 records? Perhaps I am not understanding the gist of the meaning of the orientation. Any input is appreciated.

chr1 809757 17+0- chr12 38597545 3+17- CTX -149 99 17 NA|17
chr1 809757 17+9- chr12 38597387 9+0- CTX -149 99 9 NA|9


Best,
CSoong

[pickw]

chr1 809757 17+0- chr12 38597545 3+17- CTX -149 99 17 NA|17

This means 17 plus (+) strand reads at chr1:809757 are discordantly mapped, 0 minus (-) strand reads at chr1:809757 are discordantly; whereas at chr12:38597545, 3 plus strand reads and 17 minus strand reads are discordantly mapped. Out of all these discordantly mapped reads, 17 are paired between these 2 loci, supporting this translocation. So this translocation fuses the + strand of chr1:809757 to the + strand of chr12:38597545.

chr1 809757 17+9- chr12 38597387 9+0- CTX -149 99 9 NA|9

This translocation fuses the plus strand of chr1:809757 to the negative strand of chr12:38597387.

You can post help to [email protected]
Thank.

[csoong]

Thanks pickw,

Isn't it weird that the 3rd column of the first record is 17+0 while the 3rd column of 2nd row is 17+9-?

Wouldn't both have same discordant mapped reads?

[pickw]

Indeed. Not sure why this could have happened. Which version are you using?

[csoong]

1.1 2010-10-25

[Xian]

pickw please correct me if I'm wrong.

csoong, each sv call in breakdancer is decided by paired nodes. Each node contains reads within a particular region of the sequence. A particular read may support paired node 1 (ex. if has the paired reads in node 1), or it may pair up with node 2. Whenever the read does not match any when the node containing this read is trying to match another node, it will remain for the support of other nodes.

That being said, in your case, I believe the order of the first and second line being processed should be switched since it is ordered by the CHR2 and POS2 if I remember it right. Let's say

node 1 stands for chr1 809757
node 2 stands for chr12 38597387
node 3 stands for chr12 38597545

When node 1 and node 2 try to pair with each other, it finds in node 1, there are 17+ and 9- reads not paired with any other node yet, and it find 9+ reads in node 2 that can pair with node 1. So the program happily removed those 9- reads in node 1 and 9+ reads in node 2 for registration, since they already did their job.

That's why in the first line, you wouldn't see any minus strand on node 1, since they are gone to produce the second line. What about the rest of 17+ in node 1? Well, they happily paired with node 3.

Let me know if this is not clear. Thanks,

[csoong]

Very clear, Xian. In fact, I did re-sort the output; you were right on about the order the records should appear originally.