bowtie2 outputs empty file

[leontp587]

I'm trying to align a paired reads fastq file to the hg19 genome using bowtie2 in Galaxy. The paired ends files are the output of a fastq groomer and are about 3GB each and contains reads like these:

@ERR010982.1460.2 SOLEXA-GA01_1:1:1:21:1187 length=76
AGTTATGATTTTTGTTAGTCTTTTTGTCTTATTATTCTTCCTTAGGATTATAACAACTACTCTAACCTTTTGTTCT
+ERR010982.1460.2 SOLEXA-GA01_1:1:1:21:1187 length=76
!"""!""!""""""""!"!"""""""!"""""""""""""""""""""!!"!"""!!"!!!!!!!!!!!!!!!!!!

The bowtie2 syntax as run by galaxy is:
bowtie2-build "/home/leon/ref_data/fa/hg19.fa" genome; ln -s "/home/leon/ref_data/fa/hg19.fa" genome.fa; bowtie2 -p ${GALAXY_SLOTS:-4} -x genome -1 /home/leon/galaxy-dist/database/files/000/dataset_19.dat -2 /home/leon/galaxy-dist/database/files/000/dataset_20.dat -I 0 -X 250 | samtools view -Su - | samtools sort -o - - > /home/leon/galaxy-dist/database/files/000/dataset_21.dat

For some reason, the bam file that's generated after this runs for several hours is only 62 bytes long, meaning nothing got aligned! What could I be doing wrong? This is the first time I'm aligning a genome and so could be royally screwing things up.

[westerman]

Probably better to post this to the Galaxy forum.

I am unsure which Galaxy instance you are using but my first observation is that you are trying to build the index for a commonly used genome -- hg19. Why not use the built-in index? My suspicion is that if you are using the public Galaxy instance that you are running out disk space or time when building the index.