RNA-SeQC gatkFlags option reading first word only

Question

[Jorge Amigo]

I have this issue: when I try to test RNA-SeQC using GATK's multi-threading I'm not seeing the entire gatkFlags string being read:

java -jar $HOME/ngs/software/RNA-SeQC_v1.1.8.jar \
-strictMode \
-o $RNASEQC_FOLDER \
-r $REFERENCE_FASTA \
-t $REFERENCE_TRANSCRIPTOMIC \
-s $RNASEQC_FOLDER/$SAMPLE.bams.txt \
-gatkFlags "--num_threads $THREADS"
Picked up _JAVA_OPTIONS: -Xmx1g
RNA-SeQC v1.1.8.1 07/11/14
Additional GATK flags provided: --num_threads

considering that only the first word of the string was being read then I thought that I could use -nt$THREADS, but the -n even inside the string's quotation marks seems to do something unexpected, hence an error is thrown out:

java -jar $HOME/ngs/software/RNA-SeQC_v1.1.8.jar \
-strictMode \
-o $RNASEQC_FOLDER \
-r $REFERENCE_FASTA \
-t $REFERENCE_TRANSCRIPTOMIC \
-s $RNASEQC_FOLDER/$SAMPLE.bams.txt \
-gatkFlags "-nt$THREADS"
Picked up _JAVA_OPTIONS: -Xmx1g
RNA-SeQC v1.1.8.1 07/11/14
org.apache.commons.cli.MissingArgumentException: Missing argument for option: gatkFlags

any ideas about how to override this anoying behaviour?

EDIT: just solved this issue and confirmed that the gatkFlags does accept any GATK options in a string, but not all the walkers do accept all those options. for the record, multithreading is not supported:

The analysis IntronicExpressionReadBlock currently does not support parallel execution.  Please run your analysis without the -nt option.

[Santosh Anand]

Hi Jorge, there seems to be something weird happening. I ran the same version of RNASeqC on commandline as yours , and it picked up the number of threads correctly. My actual command line is this

java8 -jar ~/bin/RNA-SeQC_v1.1.8.jar -strictMode -o tmp_out   -gatkFlags "--num_threads $THREADS" -s "test|CTRL1_S1_Aligned.sortedByCoord.out.bam|testNote"  -t trans.gtf -r /data/db/genomes/hg19.fa

and info message from RNAseqC was this

RNA-SeQC v1.1.8.1 07/11/14
Additional GATK flags provided: --num_threads 5

I had different error in running the prog, but at least it is picking the num of threads right. Are you sure that the THREADS variable is correctly set?

[Jorge Amigo]

it seems to be a problem with my java calling script. I also use aliases for java7 and java8, although I didn't include them in the original question. it works fine when calling java directly and not through the aliases, so I'll have to correct my aliases for the future. thanks anyway ;)

by the way, RNA-SeQC is based on GATK code, which was supposed to work with java7. java8 compatibility was added at GATK v3.6, but RNA-SeQC is older.

[Santosh Anand]

Happy that it is resolved. Shell is very weird and tricky sometimes!

[Santosh Anand]

Are you running the command from inside a file? Does it run fine when you run it on command line directly? Note when you run from inside a file, the shell will remove the double quotes before the rnaseqc program has any chance to parse it. So you would need to protect the quotes by adding backslashes (\). ps: your 2nd example seems to be the same as first?!

[Jorge Amigo]

I've run it both from a script and manually, both with single and double quotation marks, both using a variable $THREADS and a fixed value. no success.

I've edited the second example, since it was a copy-paste problem. I first tried using the large '--num_threads' option and then I also tried the shorter '-nt' option. no success too.

[Santosh Anand]

Try adding backslashes before both of the quotes (and first try on command line). Single quotes wont work at all because the $THREADS variable will never get expanded inside single quotes. Ps: am not sure gatk engine parses command like options like shell. So if you say -nt$THREADS, and if $THREADS is 4, the gatk will read it as -nt4 (without space) and might look for an option called -nt4, which obviously doesn't exist.

[Jorge Amigo]

I tried both escaping and not escaping the quotes with no success. I even found out that the gatkFlags option even accepts a string (well, a single word) even without being quoted. but nothing gets recognized after the first word.