GATK CalculateContamination - zeros in output

Question

[wiki97]

Hi,

I am new to exome-seq and would be grateful for any suggestions :)

I want to run GATK CalculateContamination (GATK 4.1.8.1), before calling variants with MuTect2. CalculateContamination tool returns "SUCCESS" message, but with warnings, and I get only "0" values in my output file :

sample  contamination   error
my_sample  0.0          0.0

---

Here are warnings after running GATK CalculateContamination:

WARN  KernelSegmenter - Number of points needed to calculate local changepoint costs (2 * window size = 100) exceeds number of data points (12).  Local changepoint costs will not be calculated for this window size.
WARN  KernelSegmenter - No changepoint candidates were found.  The specified window sizes may be inappropriate, or there may be insufficient data points.
INFO  KernelSegmenter - Found 0 changepoints after applying the changepoint penalty.
WARN  KernelSegmenter - Specified dimension of the kernel approximation (100) exceeds the number of data points (12) to segment; using all data points to calculate kernel matrix.

---

Here is my code:

gatk CalculateContamination \
-I GetPileupSummaries_output.table \
-O CalculateContamination_output.contamination.table

I searched GATK documentation but didn't find any instructions there. I will be most grateful for any help :)

Answer 1

[DavidStreid]

If you used GATK GetPileupSummaries, did you use a VCF/interval list file that would have positions your input BAM would cover?

[wiki97]

Yes, I used GATK GetPileupSummaries and I used an interval file, this file was output from GATK RealignerTargetCreator tool.

Answer 2

[aldhairmedico]

Hi @wiki97, I had the same issue using GATK 4.2:

Using BWA_MEM as aligner I get the:

WARN KernelSegmenter - Number of points needed to calculate local changepoint costs (2 * window size = 100) exceeds number of data points (12). Local changepoint costs will not be calculated for this window size. WARN KernelSegmenter - No changepoint candidates were found. The specified window sizes may be inappropriate, or there may be insufficient data points. INFO KernelSegmenter - Found 0 changepoints after applying the changepoint penalty. WARN KernelSegmenter - Specified dimension of the kernel approximation (100) exceeds the number of data points (12) to segment; using all data points to calculate kernel matrix. Followed by a SUCCESS message and plausible values

Using Bowtie2 as aligner I get: No warnings but the .contamination tables are

sample contamination error my_sample
1A 0.0 0.0

I don't really know why is happening this. Does anyone know if we can skip this part using GATK?