def align_reads_bowtie2(unaligned_bam_file,
index_base,
alignment_file,
temp_prefix,
num_threads=1):
"""Aligns unaligned bam file."""
bowtie2_align_parameter = (
'--preserve-tags -D 20 -R 3 -N 1 -L 15 -i S,1,0.50')
bowtie2_align_parameter = ' '.join(
['-p', str(num_threads), bowtie2_align_parameter])
cmd = [
get_binary_path(binary_name='bowtie2'), bowtie2_align_parameter,
' '.join(['-x', str(index_base), '-b',
str(unaligned_bam_file)]), '|',
get_binary_path(binary_name='samtools'), 'view -uS - | ',
get_binary_path(binary_name='samtools'), 'sort -T', temp_prefix, '-@',
str(num_threads), '-o',
str(alignment_file), '-'
]
_, errs = run_executable(cmd_line=cmd)
return alignment_file, errs
def count_bus_file(bus_file,
output_directory,
output_prefix,
t2g_file,
ec_file,
transcripts_file):
"""Generates matrix."""
cmd = [
get_binary_path(binary_name='bustools'),
'count',
'-o',
str(output_prefix),
'--genecounts',
'-g',
str(t2g_file),
'-e',
str(ec_file),
'-t',
str(transcripts_file),
str(bus_file)
]
outs, errs = run_executable(cmd)
return output_prefix
def correct_cell_barcodes(cb_file,
output_directory,
bus_file,
corrected_bus_file):
"""Corrects cell barcodes."""
CELL_BARCODE_FILE = output_directory / 'barcodes_no_suffix.tsv'
with open_by_suffix(file_name=str(cb_file), mode='r') as f:
with open_by_suffix(
file_name=str(CELL_BARCODE_FILE),
mode='w') as fo:
for line in f:
i = line.rstrip().split('-')[0]
fo.write(i + '\n')
logger.info('Number of whitelisted cell barcodes: '
+ f'{len([i for i in open(CELL_BARCODE_FILE)])}')
cmd = [
get_binary_path(binary_name='bustools'),
'correct',
'-w',
str(output_directory / 'barcodes_no_suffix.tsv'),
'-o',
str(corrected_bus_file),
str(bus_file)
]
outs, errs = run_executable(cmd)
logger.info(errs)
return corrected_bus_file
def align_reads_kallisto(read1_file,
read2_file,
kallisto_index,
output_directory,
technology,
num_threads=1):
"""Aligns reads."""
cmd = [
get_binary_path(binary_name='kallisto'),
'bus',
'-i',
str(kallisto_index),
'-o',
str(output_directory),
'-x',
technology,
'-t',
str(num_threads),
'-n',
read1_file,
read2_file
]
outs, errs = run_executable(cmd)
logger.debug(' '.join(cmd))
return outs, errs
def build_kallisto_index(kallisto_index,
kmer,
fasta_file):
"""Builds kallisto index.
A wrapper of `kallisto index [arguments] FASTA-files`.
Parameters
----------
kallisto_index : str
The path and name of kallisto index.
kmer : int
k-mer length, odd number.
fasta_file : str
The path and name of generated fasta file.
Returns
-------
str
The path and name of generated kallisto index.
"""
cmd = [
get_binary_path(binary_name='kallisto'),
'index',
'-i',
str(kallisto_index),
'-k',
str(kmer),
str(fasta_file)
]
outs, errs = run_executable(cmd_line=cmd)
logger.info(errs)
return kallisto_index
def build_bwa_index(fasta_file):
"""Builds bwa index."""
cmd = [get_binary_path(binary_name='bwa'), 'index', str(fasta_file)]
cmd = ' '.join(cmd)
_, errs = run_executable(cmd_line=cmd, use_shell=True)
return fasta_file, errs
def build_bt2_index(fasta_file, index_base):
"""Builds bowtie2 index."""
cmd = [
get_binary_path(binary_name='bowtie2-build'),
str(fasta_file),
str(index_base)
]
_, errs = run_executable(cmd_line=cmd)
return index_base, errs
def align_reads_bwa(modified_read_file,
index_base,
alignment_file,
temp_prefix,
num_threads=1):
"""Aligns modified fastq file."""
bwa_align_parameter = ['-t', str(num_threads), '-C']
samtools_path = get_binary_path(binary_name='samtools')
cmd = [
get_binary_path(binary_name='bwa'), 'mem',
' '.join(bwa_align_parameter),
str(index_base),
str(modified_read_file), '|', samtools_path, 'view -uS - |',
samtools_path, 'sort -T', temp_prefix, '-@',
str(num_threads), '-o',
str(alignment_file), '-'
]
cmd = ' '.join(cmd)
_, errs = run_executable(cmd_line=cmd, use_shell=True)
return alignment_file, errs
def sort_bus_file(bus_file,
sorted_bus_file,
num_threads):
"""Sorts bus file."""
# logger.info(f'Number of threads: {num_threads}')
cmd = [
get_binary_path(binary_name='bustools'),
'sort',
'-t',
str(num_threads),
'-o',
str(sorted_bus_file),
str(bus_file)
]
outs, errs = run_executable(cmd)
logger.info(errs)
return sorted_bus_file