MarkDuplicates not working on CG BAM files?

[biscuit13161]

Hi everyone,

I've created some BAM files from some Complete Genomics data using their conversion script so I can run it through a GATK/Picard pipeline and everything seems to be running sweet until I get to run the IndelRealigner. I seem to keep getting a duplicate record error from IndelRealigner, whether I use MarkDuplicates with REMOVE_DUPLICATE = true or false.

java -Xmx12g -Xms12g -jar ~/apps/picard/MarkDuplicates.jar INPUT=recal_GS14972.bam OUTPUT=GS14972_dedup_test.bam METRICS_FILE=GS14972_dedup_test.metrics REMOVE_DUPLICATES=false CREATE_INDEX=true CREATE_MD5_FILE=true TMP_dir=~/recal_GS14972

##### ERROR ------------------------------------------------------------------------------------------
##### ERROR A USER ERROR has occurred (version 2.3-9-ge5ebf34):
##### ERROR The invalid arguments or inputs must be corrected before the GATK can proceed
##### ERROR Please do not post this error to the GATK forum
##### ERROR
##### ERROR See the documentation (rerun with -h) for this tool to view allowable command-line arguments.
##### ERROR Visit our website and forum for extensive documentation and answers to
##### ERROR commonly asked questions http://www.broadinstitute.org/gatk
##### ERROR
##### ERROR MESSAGE: Error caching SAM record GS53717-FS3-L03-15:16450476, which is usually caused by malformed SAM/BAM files in which multiple identical copies of a read are present.
##### ERROR ------------------------------------------------------------------------------------------

There are clear duplicates in the input files, but it appears that MarkDuplicates is not finding them (the metrics file shows no found duplicates). Looking by eye I can only see differences between the reads at ZA and ZI columns, but I've been unable to identify their meaning.

I'd really appreciate any thoughts or comments that would help me progress with this analysis.

Thanks.