limitation in command length

naarkhoo

I wonder if there is any limitation for the command length. I provide more than 90 bam files, including their directory address to GATK for variant calling using UnifiedGenotyper, and it seems it doesn't parse whole the command. I know It could be because of java, also. Have you ever experiences similar problem ? Is there any way to provide a directory path instead of all single bam files path as the input ? thanks,

Geraldine_VdAuwera

Hi there,

You can input your bam files by simples writing them in a text file (one filepath per line) and passing that bam list file with the -I argument. For example, let's say I have this list of bamfiles (they can be in different directories):

/my_data_directory/my_bamfile_1.bam
/my_data_directory/my_bamfile_2.bam
/my_data_directory/my_bamfile_3.bam
/my_other_data_directory/my_bamfile_4.bam
/my_other_data_directory/my_bamfile_5.bam

I just write those lines in a text file that I call my_bam_files.list and add it to my command like this:

java -jar GenomeAnalysisTK.jar -T UnifiedGenotyper -R my_reference.fasta -I my_bam_files.list --dbsnp dbSNP.vcf -o snps.raw.vcf 

You can do that with any tool when you want to pass a list of bam files rather than write them all out at the command line.

naarkhoo

I don't know if it is relevant but it complains about un-index bam files. Should all bai files (indeces) located in the same directory ? is there option to set the path of bai files also?

naarkhoo

Or to know which of the listed BAM files is not index-ed

Geraldine_VdAuwera

Yes, the bai file indices should be in the same directory as the bams they index. There is no option to provide separate paths for the bam and bai.