Error in CollectGcBiasMetrics during ATAQC module

Hello,

I am running the pipeline starting with nodup.bam files. The pipeline seemed to run successfully with previous version of the pipeline. However, when I downloaded the latest version of the pipeline couple of days ago, the pipeline breaks at picard’s CollectGcBiasMetrics command during ATAQC module with Exception in thread "main" java.lang.ArrayIndexOutOfBoundsException: Index 130695116 out of bounds for length 130694994 error. After some investigation, I found where the out of bounds numbers are coming from. The length of chr10 is 130694993 and there is a read that maps to chr2 at 130695116. What is not clear is why picard tries to use length of chr10 for read mapping to chr2. I did little more investigation to find out the scripts that are calling picard’s CollectGcBiasMetrics. It looks like before the get_gc function, which runs CollectGcBiasMetrics, is called by encode_ataqc.py, the input bam file is processed to remove read groups. However, this feature seem to be missing in older version of encode_ataqc.py. I am not sure if this is the culprit or I am doing something wrong. Please advice.

Thank you for your time.

debug_bfadbaaa-2396-4275-b11f-0b1e5294afab.tar.gz

[leepc12]

Yes, you can keep using the TSV file mm10.tsv. In your input JSON file, define the following to override reference FASTA defined in the TSV file.

{
    "atac.ref_fa" : "PATH_FOR_YOUR_OWN_FASTA"
}

[ravipatel4]

That worked! Thanks @leepc12 and @vervacity.