def trimmomatic_se(fastq1,
crop_length,
crop_length_tol,
out_dir,
nth=1,
java_heap=None):
prefix = os.path.join(out_dir, os.path.basename(strip_ext_fastq(fastq1)))
crop_length_tol = abs(crop_length_tol)
min_length = crop_length - crop_length_tol
cropped = '{p}.crop_{cl}-{tol}bp.fastq.gz'.format(p=prefix,
cl=crop_length,
tol=crop_length_tol)
if java_heap is None:
java_heap_param = '-Xmx6G'
else:
java_heap_param = '-Xmx{}'.format(java_heap)
cmd = 'java -XX:ParallelGCThreads=1 {param} -jar {jar} SE -threads {nth} '
cmd += '{fq1} {cropped} MINLEN:{ml} CROP:{cl}'
cmd = cmd.format(param=java_heap_param,
jar=locate_trimmomatic(),
nth=nth,
fq1=fastq1,
cropped=cropped,
ml=min_length,
cl=crop_length)
run_shell_cmd(cmd)
return cropped
def trimmomatic_pe(fastq1,
fastq2,
crop_length,
out_dir_R1,
out_dir_R2,
nth=1,
java_heap=None):
prefix_R1 = os.path.join(out_dir_R1,
os.path.basename(strip_ext_fastq(fastq1)))
prefix_R2 = os.path.join(out_dir_R2,
os.path.basename(strip_ext_fastq(fastq2)))
cropped_R1 = '{}.crop_{}bp.fastq.gz'.format(prefix_R1, crop_length)
cropped_R2 = '{}.crop_{}bp.fastq.gz'.format(prefix_R2, crop_length)
tmp_cropped_R1 = '{}.tmp'.format(cropped_R1)
tmp_cropped_R2 = '{}.tmp'.format(cropped_R2)
if java_heap is None:
java_heap_param = '-Xmx6G'
else:
java_heap_param = '-Xmx{}'.format(java_heap)
cmd = 'java -XX:ParallelGCThreads=1 {} -jar {} PE -threads {} '
cmd += '{} {} {} {} {} {} MINLEN:{} CROP:{}'
cmd = cmd.format(java_heap_param, locate_trimmomatic(), nth, fastq1,
fastq2, cropped_R1, tmp_cropped_R1, cropped_R2,
tmp_cropped_R2, crop_length, crop_length)
run_shell_cmd(cmd)
rm_f([tmp_cropped_R1, tmp_cropped_R2])
return cropped_R1, cropped_R2
def trimmomatic_pe(fastq1,
fastq2,
crop_length,
crop_length_tol,
out_dir_R1,
out_dir_R2,
nth=1,
java_heap=None):
prefix_R1 = os.path.join(out_dir_R1,
os.path.basename(strip_ext_fastq(fastq1)))
prefix_R2 = os.path.join(out_dir_R2,
os.path.basename(strip_ext_fastq(fastq2)))
crop_length_tol = abs(crop_length_tol)
min_length = crop_length - crop_length_tol
cropped_R1 = '{p}.crop_{cl}-{tol}bp.fastq.gz'.format(p=prefix_R1,
cl=crop_length,
tol=crop_length_tol)
cropped_R2 = '{p}.crop_{cl}-{tol}bp.fastq.gz'.format(p=prefix_R2,
cl=crop_length,
tol=crop_length_tol)
tmp_cropped_R1 = '{}.tmp'.format(cropped_R1)
tmp_cropped_R2 = '{}.tmp'.format(cropped_R2)
if java_heap is None:
java_heap_param = '-Xmx6G'
else:
java_heap_param = '-Xmx{}'.format(java_heap)
cmd = 'java -XX:ParallelGCThreads=1 {param} -jar {jar} PE -threads {nth} '
cmd += '{fq1} {fq2} {cropped1} {tmp_cropped1} {cropped2} {tmp_cropped2} '
cmd += 'MINLEN:{ml} CROP:{cl}'
cmd = cmd.format(param=java_heap_param,
jar=locate_trimmomatic(),
nth=nth,
fq1=fastq1,
fq2=fastq2,
cropped1=cropped_R1,
tmp_cropped1=tmp_cropped_R1,
cropped2=cropped_R2,
tmp_cropped2=tmp_cropped_R2,
ml=min_length,
cl=crop_length)
run_shell_cmd(cmd)
rm_f([tmp_cropped_R1, tmp_cropped_R2])
return cropped_R1, cropped_R2
def trimmomatic_se(fastq1, crop_length, out_dir, nth=1, java_heap=None):
prefix = os.path.join(out_dir, os.path.basename(strip_ext_fastq(fastq1)))
cropped = '{}.crop_{}bp.fastq.gz'.format(prefix, crop_length)
if java_heap is None:
java_heap_param = '-Xmx6G'
else:
java_heap_param = '-Xmx{}'.format(java_heap)
cmd = 'java -XX:ParallelGCThreads=1 {} -jar {} SE -threads {} '
cmd += '{} {} MINLEN:{} CROP:{}'
cmd = cmd.format(java_heap_param, locate_trimmomatic(), nth, fastq1,
cropped, crop_length, crop_length)
run_shell_cmd(cmd)
return cropped
def trimmomatic_se(fastq1,
crop_length,
crop_length_tol,
phred_score_format,
out_dir,
nth=1,
java_heap=None):
prefix = os.path.join(out_dir, os.path.basename(strip_ext_fastq(fastq1)))
crop_length_tol = abs(crop_length_tol)
min_length = crop_length - crop_length_tol
cropped = '{p}.crop_{cl}-{tol}bp.fastq.gz'.format(p=prefix,
cl=crop_length,
tol=crop_length_tol)
if java_heap is None:
java_heap_param = '-Xmx6G'
else:
java_heap_param = '-Xmx{}'.format(java_heap)
phred_score_format = phred_score_format.lower()
if phred_score_format == 'auto':
phred_score_param = ''
elif phred_score_format == 'phred33':
phred_score_param = '-phred33'
elif phred_score_format == 'phred64':
phred_score_param = '-phred64'
else:
raise ValueError('Wrong phred_score_format!')
cmd = 'java -XX:ParallelGCThreads=1 {param} -jar {jar} SE -threads {nth} {phred_score_param} ' \
'{fq1} {cropped} MINLEN:{ml} CROP:{cl}'.format(
param=java_heap_param,
jar=locate_trimmomatic(),
nth=nth,
phred_score_param=phred_score_param,
fq1=fastq1,
cropped=cropped,
ml=min_length,
cl=crop_length,
)
run_shell_cmd(cmd)
return cropped