ModelSegments output from Somatic CNV

Follow

Indrani Datta

• April 29, 2024 13:44

Dear GATK forum,

I have performed somatic CNV analysis jointly with coverage data with ModelSegments with GATK 4.4.0.0 with hg38 with regular parameters and output from this run leads to figure 1 which shows hypersegmentation and some of the segments are log2 copy ratio is positive and while some of them are negative.

gatk --java-options "-Xmx7g" ModelSegments \
    --denoised-copy-ratios tumor.denoisedCR.tsv \
    --output sandbox  \
    --output-prefix Tumor_model_segment

 

So, I changed some of the parameters in modelsegment such as shown below , now hypersegmentation is resolved but the output only have few segments and all of them are deletion such as segment MEAN_LOG2_COPY_RATIO are all negative. How is it possible that all of them are negative after changing the parameters. 

Please suggest.

gatk --java-options "-Xmx7g" ModelSegments \
    --denoised-copy-ratios tumor.denoisedCR.tsv \
    --number-of-changepoints-penalty-factor 5.0 \
    --kernel-variance-copy-ratio 5.0 \
    --kernel-variance-allele-fraction 5.0 \
    --output sandbox  \
    --output-prefix model_segment_output

0

• 

  Tamar Melman

  • May 13, 2024 14:32

  Those are indeed very low numbers, insufficient for depth-based CNV calling. Double check that you supplied the targeted intervals.

  0

  Comment actions Permalink
• 

  Indrani Datta

  • May 14, 2024 17:09

  Thank you for your reply. The interval list is comparable to the library which was profiled with the samples.

  Best.

  0

  Comment actions Permalink

Please sign in to leave a comment.