JointGermlineCNVSegmentation (BETA)

Follow

Stephen Shoebridge

• July 03, 2024 10:46

a) GATK version used: 4.5
b) Exact command used: 

    JointGermlineCNVSegmentation \

    --reference /mnt/ref/$bwa_reference \

    --arguments_file /mnt/output/vcf_seg_samples.txt \

    --model-call-intervals /mnt/output/filtered_intervals.interval_list \

    --pedigree /mnt/output/vtc.ped \

    --output /mnt/output/cnv_seg_clustered.vcf.gz

Subject:

Could you briefly describe how the merging of segments works as we have two segments that overlap with one being slightly longer, but they differ in DEL/DUP calls (see below). We are just confused how an addition of a small interval at the end is shifting the calling and how best we can interepret this result.

Many Thanks,

Stephen

chr1    109687517   CNV_chr1_109687517_109690964    T   <DEL>   .   .  

chr1    109687517   CNV_chr1_109687517_109693620    T   <DEL>,<DUP>

 

0

• 

  Gökalp Çelik

  • July 04, 2024 12:18

  Hi Stephen Shoebridge

  JointGermlineCNVSegmentation requires overlapping segments to be 80% overlapping in length to cluster them together. Your sites are only about 56 percent overlapping which puts these 2 variants in different clusters. 

  If you wish to modify the overlapping percentage value you may wish to modify the below parameter 

  --clustering-interval-overlap <Double>
                                Minimum interval reciprocal overlap for clustering  Default value: 0.8.

  I hope this helps. 

   

  0

  Comment actions Permalink
• 

  Stephen Shoebridge

  • July 04, 2024 12:21

  Thanks,

  Definitetly looking at the clustering may be a the means to calibrating our results. Thanks for the quick response.

  Best,

  Stephen

  0

  Comment actions Permalink

Please sign in to leave a comment.