star-reference.qsubsec

# This template generates a salmon index from a given reference set.
# Alastair Droop, 2018-11-21

# Get the necessary Python functions:
from os import makedirs
from os.path import expanduser, expandvars

section('STAR-INDEX-{REF_ID}', description='Generate a Salmon index ({INDEX_PREFIX}-{REF_ID}) from genome reference set {REF_ID}')
if('{USE_WHOLE_NODE}' == 'yes'):
    limits(h_rt='{MAX_TIME}', nodes=1)
else:
    limits(h_rt='{MAX_TIME}', h_vmem='{MAX_MEM}')
    options('pe smp {MAX_THREADS}')
options('V', 'cwd', 'notify')

# Attempt to set up the reference folder:
try:
    makedirs(expandvars(expanduser('{LOG_DIR}')))
    message('created reference log directory "{LOG_DIR}"')
except FileExistsError: message('reference log directory "{LOG_DIR}" already exists')
except: raise Exception('Failed to create reference log directory {LOG_DIR}')
outputs('{LOG_DIR}')

# Check that the directories are all OK:
for p in ['{INDEX_DIR}', '{LOG_DIR}', '{TEMP_DIR}']:
    require(p, 'PATH_READABLE')
    require(p, 'PATH_WRITABLE')

# Check that all of the executables are OK:
require('{STAR_EXEC}', 'PATH_EXECUTABLE')

# Submit the salmon index command:
command('{STAR_EXEC} --runMode genomeGenerate --runThreadN {MAX_THREADS} --limitGenomeGenerateRAM {MAX_STAR_MEM} --outTmpDir {TEMP_DIR}/STAR --genomeDir {INDEX_DIR} --genomeFastaFiles {REFSET_FASTA} --sjdbGTFfile {REFSET_GTF} --sjdbOverhang {SJ_OVERHANG} --outFileNamePrefix {LOG_DIR}/', name='generate-index', log=True, test=True)

# Add a link to the original reference FASTA & GTF files:
command('ln -s {REFSET_FASTA} {INDEX_DIR}/{INDEX_PREFIX}-{REF_ID}.fasta', name='ln_fasta')
command('ln -s {REFSET_GTF} {INDEX_DIR}/{INDEX_PREFIX}-{REF_ID}.gtf', name='ln_gtf')