VarScan2 Parsing Error

Question

[ATpoint]

A simple question: Did anyone ever encountered this error when using VarScan2 processSomatic, to extract somatic calls from raw VarScan2 output:

Warning: Unable to parse chrom/position from chr1 94134 . C T . PASS DP=33;SS=1;SSC=0;GPV=1,042E-3;SPV=9,3168E-1 GT:GQ:DP:RD:AD:FREQ:DP0/1:.:10:6:4:40%:6,0,1,3 0/1:.:23:18:5:21,74%:18,0,5,0

This command line:

$SAMTOOLS mpileup -q 20 -Q 25 -B -f $HG38 $RMDUP_DIR/${BASENAME}-n_rmdup.bam $RMDUP_DIR/${BASENAME}-t_rmdup.bam | $VARSCAN somatic stdin ${BASENAME} -mpileup --strand-filter 1 --output-vcf && $VARSCAN processSomatic ./processSomatic/${BASENAME}.snp --max-normal-freq 0.01

The exact same command was used multiple times before, even on the same files, but this never occurred. Since today, when processing previously used files and newly generated ones, this error pops up. Googling the error message only directed me to the source code of VS on github, and a lonely post on seqanswers, describing the same problem without a solution. Any ideas?

EDIT: Adding export LC_ALL=en_US.UTF-8 to the beginning of the script solved the issue. Seemed to be an issue with my german locale, as Pierre suggested.

[Pierre Lindenbaum]

can you please show us the VCF header for INFO and FORMAT please ?

[ATpoint]

Here is the header section.

Answer 1

[Pierre Lindenbaum]

One of your error could be

SPV=9,3168E-1

while it should be

SPV=9.3168E-1

may be your using a foreign locale ? try to set your java encoding as UTF-8 .

java -Dfile.encoding=UTF-8 -jar varscan.jar (...)

[ATpoint]

Great, thanks for the hint with the locale. Even though your suggestion did not solved it right away, simply adding export LC_ALL=en_US.UTF-8 to the beginning of the script without further changes on the java command solved the issue.