Pindel predicts inversions using only one side of the anchors

[choijae3]

Hi folks

I have a somewhat unique problem to detect SV (especially inversions) using PE data. I've been using pindel to detect SV. Without going too much in detail of my genome of interest it's a size of about 1.2mb and I have very good reason to believe that the whole region is perhaps evolutionary close to being a pseudogene. Thus there are alot of mutations and SV segregating in the genome.
When I use pindel to detect inversions interestingly all the inversion break points I find are from one side with pretty good coverage (~20X reads predict the breakpoint). I have a hard time understanding why only from one side it predicts the breakpoint since if there are 20 reads predicting one side there should be also many reads on the other side predicting the break point as well.
For example

####################################################################################################
0 INV 7209 NT 0:0 "":"" ChrID genome BP 426826 434036 BP_range 426826 434036 Supports 21 19 + 0 0 - 21 19 S1 22 SUM_MS 1260 1 NumSupSamples 1 1 lib 0 0 21 19

All the INV I find are like that where only one side of the break point is found. I'm just wondering what kind of phenomenon could lead to such results.

Thanks!