def reads_mapping(sortmerna_bin,
fasta_ref_path,
index_ref_basepath,
reads_path,
output_basepath,
best,
min_lis,
evalue,
cpu,
verbose=False):
logger.info('--- Reads mapping against scaffolds ---')
logger.debug('Database:%s' % index_ref_basepath)
parameters = {
'fasta_ref': fasta_ref_path,
'index_ref': index_ref_basepath,
'reads': reads_path,
'output': output_basepath,
'best': best,
'min': min_lis,
'evalue': evalue,
'cpu': cpu
}
cmd_line = '{bin} --ref {fasta_ref},{index_ref} --reads {reads} --aligned {output}' \
' --fastx --sam --blast "1" --log --best {best} --min_lis {min}' \
' -e {evalue:.2e} -a {cpu}'.format(bin=sortmerna_bin, **parameters)
if verbose:
cmd_line += ' -v'
runner.logged_check_call(cmd_line, verbose=verbose)
def run(self):
"""
Check I/O and simply run the command
"""
if self.cmd_line is None:
logger.fatal('You have to call "build_command_line" method before "run" method')
sys.exit("No command line available")
if self.fastq_file is None or not os.path.isfile(self.fastq_file):
logger.fatal('The input reads file does not exists:%s' % self.fastq_file)
sys.exit("Can't assemble %s" % self.fastq_file)
if self.workdir is None:
logger.debug("Workdir is not set")
sys.exit("Can't assemble %s" % self.fastq_file)
elif os.path.isdir(self.workdir):
logger.debug("Remove previous assembler working dir before assembling:%s" % self.workdir)
shutil.rmtree(self.workdir)
os.mkdir(self.workdir)
runner.logged_check_call(self.cmd_line)
return self.fasta_file
def make_krona_plot(krona_bin, krona_text_file, krona_html_file):
logger.info('Make krona plot')
logger.debug('Krona tab file:%s' % krona_text_file)
cmd_line = '{bin} {txt} -o {html}'.format(bin=krona_bin,
txt=krona_text_file,
html=krona_html_file)
runner.logged_check_call(cmd_line)
def index_ref(indexdb_bin_path, input_fasta_ref_path, output_basepath, max_mem, verbose=False):
logger.info('--- Indexing scaffolds ---')
logger.debug('File to index: %s' % input_fasta_ref_path)
parameters = { 'input': input_fasta_ref_path, 'output': output_basepath, 'max_mem': max_mem }
cmd_line = '{bin} --ref {input},{output} -m {max_mem}'.format(bin=indexdb_bin_path, **parameters)
if verbose:
cmd_line += ' -v'
runner.logged_check_call(cmd_line, verbose=verbose)
def get_best_matches(best_matches_bin, input_blast_path, out_blast_path, max_mem, cpu):
outdir = os.path.dirname(out_blast_path)
sort_bin = 'sort -T ' + outdir + ' -S ' + str(max_mem)
sort_bin += 'M --parallel ' + str(cpu)
cmd_line = sort_bin + ' -k1,1V -k12,12nr ' + input_blast_path
cmd_line += ' | ' + best_matches_bin + ' -p 0.99 -o ' + out_blast_path
runner.logged_check_call(cmd_line)
def run_rdp_classifier(rdp_exe,
in_fasta,
out_classification_file,
cutoff=0.8,
gene='16srrna'):
parameters = {
'fa': in_fasta,
'out': out_classification_file,
'cutoff': cutoff,
'gene': gene
}
cmd_line = '{rdp_exe} classify -c {cutoff} -f fixrank -g {gene} -o {out} {fa}'.format(
rdp_exe=rdp_exe, **parameters)
runner.logged_check_call(cmd_line)