Holiday Notice:
The Frontline Support team will be offline February 18 for President's Day but will be back February 19th. Thank you for your patience as we get to all of your questions!

PlotDenoisedCopyRatios (v4.0.1.2) error "need finite 'ylim' values"

stefan_stojanovicstefan_stojanovic Member
edited February 4 in Ask the GATK team
Hello, I'm getting the following error from PlotDenoisedCopyRatios tool, version 4.0.1.2.

> 2019-01-31T09:56:23.782315172Z Stderr: Error in plot.window(...) : need finite 'ylim' values
> 2019-01-31T09:56:23.782317446Z Calls: source ... SetUpPlot -> plot -> plot.default -> localWindow -> plot.window

I assume this is to do with the inputs denoised and standardized copy ratios, which are produced by DenoiseReadCounts (v4.0.1.2 as well). Below is chunk of these files. There are "Infinity" values in LOG2_COPY_RATIOS column so I assume this is causing PlotDenoisedCopyRatios to fail, but maybe I am wrong. There is also number of NaN values in this column.

...
CONTIG START END LOG2_COPY_RATIO
chr22 50776501 50776750 0.336554
chr22 50776751 50777000 Infinity
chr22 50777751 50778000 0.250825

chr22 50807751 50808000 NaN
chr22 50808001 50808250 NaN
chr22 50808251 50808500 NaN
...

I ran both tools with default parameter values, here are command lines.

gatk --java-options -Xmx13000m PlotDenoisedCopyRatios --output . --output-prefix cnv_NORMAL --denoised-copy-ratios cnv_NORMAL.denoisedCR.tsv --sequence-dictionary Homo_sapiens_assembly38.dict --standardized-copy-ratios cnv_NORMAL.standardizedCR.tsv

gatk --java-options -Xmx13000m DenoiseReadCounts --input cnv_NORMAL.counts.hdf5 --denoised-copy-ratios cnv_NORMAL.denoisedCR.tsv --standardized-copy-ratios cnv_NORMAL.standardizedCR.tsv --count-panel-of-normals PoN_4.0_WGS_for_public.pon.hdf5

My question is how do I solve this issue. Do I need to change some parameters or is this issue with input files?
Post edited by bshifaw on

Best Answers

Answers

  • Hi @bshifaw, I managed to find a workaround which is to use larger interval bins to collect read counts (bin_length=1000). I am still getting the same error with smaller bins (bin_length=250) which are used in best practice FullSomaticPipeline example (github.com/gatk-workflows/gatk4-somatic-with-preprocessing). It is possible my BAM files are not suitable for CNV calling (they are around 2GB each), but DenoiseReadCounts finished without errors so I'm not sure. Hope this helps your team.
Sign In or Register to comment.