CNV not called in some samples

Follow

Kate Megquier

• March 31, 2020 17:14

Dear GATK Team,

 

We've noticed a CNV being called in some samples, but not being called as significant in others, despite a mean log2 copy ratio of -26, and are unsure why this is happening.

 

We called variants on unpaired canine cancer cell lines, using a panel of normals from unmatched dogs. For calling, we used the Terra GATK somatic CNV pipeline:

• 1_CNV_Somatic_Panel (v.1.3.1)
• 2_CNV_Somatic_Pair (v.1.3.0)

 

Our data was fairly noisy with the default settings, so we ended up implementing some smoothing options:

• num_changepoints_penalty_factor 5
• kernel_variance_allele_fraction 0.8
• kernel_variance_copy_ratio 0.8

 

An example of the CNV being called in one sample but not another:

Sample 1:

chrX    position1        position2        427     -26.224015      -

 

Sample 2:

chrX    position1        position2        427     -26.003399      0

 

These examples were the same sex, and so were run with the same PoN. Can someone help us to understand why this is happening? 

 

Thank you for your help!

Best,

Kate

0

• 

  Bhanu Gandham

  • April 03, 2020 17:28

  Hi Kate Megquier

   

  Have you had a chance to look at these docs: https://gatk.broadinstitute.org/hc/en-us/articles/360035531092--How-to-part-I-Sensitively-detect-copy-ratio-alterations-and-allelic-segments

  0

  Comment actions Permalink
• 

  Kate Megquier

  • April 05, 2020 22:07

  Dear Bhanu Gandham,

  Thank you for your reply! Yes, we have read through the tutorial docs and found them very helpful, but were still unsure of why a CNV might be called significant in one sample and not in another when both had the same mean log2 copy ratio. We wondered if it had to do with the variance of the segments immediately around the CNV in question, but aren't sure how the tool calculates whether a CNV is significant.

  We also did note that the allelic segmentation is known to be less reliable when running with case-only samples, but it seemed like the document stopped short of saying that they shouldn't be used in case only studies. Does allelic segmentation add information in case-only calling, or should we be performing only copy-ratio segmentation?

  Best,

  Kate

  0

  Comment actions Permalink
• 

  Bhanu Gandham

  • April 07, 2020 20:01

  Hi Kate, 

   

  I am checking with the team and will get back to you shortly.

  0

  Comment actions Permalink
• 

  Bhanu Gandham

  • April 08, 2020 21:21

  Hi Kate,

  1. The first thing to try is to  look at the documentation for CallCopyRatioSegments https://gatk.broadinstitute.org/hc/en-us/articles/360041850791-CallCopyRatioSegments and see if changing threshold parameters fixes things.
  2. If not we would probably need to see more data, including plots and full modeled segments output and called segment outputs.
  3. Regarding the other question you asked: "doing copy-ratio + allelic segmentation improves the segmentation over just using the copy ratios (even in case only), but the allelic data is otherwise not used in CallCopyRatioSegments"
  0

  Comment actions Permalink

Please sign in to leave a comment.