def create_housekeeping_file(chr_lengths, max_points, root_dir, output_dir, logger):
max_ideograms = len(chr_lengths.keys())
template_fpath = None
circos_bin_fpath = get_path_to_program('circos')
if circos_bin_fpath:
circos_dirpath = dirname(realpath(get_path_to_program('circos')))
template_fpath = join(circos_dirpath, '..', 'libexec', 'etc', 'housekeeping.conf')
if not is_non_empty_file(template_fpath):
template_fpath = join(circos_dirpath, '..', 'etc', 'housekeeping.conf')
if not is_non_empty_file(template_fpath):
if not get_path_to_program('circos'):
msg = 'Circos is not found.'
else:
msg = 'File etc/housekeeping.conf is not found.'
logger.warning(msg + ' You will have to manually edit etc/housekeeping.conf: '
'set max_points_per_track to ' + str(max_points) + ' and max_ideograms to ' + str(max_ideograms))
return '<<include %s>>\n' % join('etc', 'housekeeping.conf')
housekeeping_fpath = join(output_dir, 'housekeeping.conf')
with open(template_fpath) as f:
with open(housekeeping_fpath, 'w') as out_f:
for line in f:
if 'max_points_per_track' in line:
out_f.write('max_points_per_track = %d\n' % max_points)
elif 'max_ideograms' in line:
out_f.write('max_ideograms = %d\n' % max_ideograms)
else:
out_f.write(line)
return '<<include %s>>\n' % relpath(housekeeping_fpath, root_dir)
def compile_glimmer(logger, only_clean=False):
tool_dirpath = os.path.join(qconfig.LIBS_LOCATION, 'glimmer')
tool_src_dirpath = os.path.join(tool_dirpath, 'src')
if not get_path_to_program('glimmerhmm', tool_dirpath):
compile_tool('GlimmerHMM', tool_src_dirpath, ['../glimmerhmm'], logger=logger, only_clean=only_clean)
return get_path_to_program('glimmerhmm', tool_dirpath)
def create_housekeeping_file(chr_lengths, max_points, root_dir, output_dir, logger):
max_ideograms = len(chr_lengths.keys())
template_fpath = None
circos_bin_fpath = get_path_to_program('circos')
if circos_bin_fpath:
circos_dirpath = dirname(realpath(get_path_to_program('circos')))
template_fpath = join(circos_dirpath, '..', 'libexec', 'etc', 'housekeeping.conf')
if not is_non_empty_file(template_fpath):
template_fpath = join(circos_dirpath, '..', 'etc', 'housekeeping.conf')
if not is_non_empty_file(template_fpath):
if not get_path_to_program('circos'):
msg = 'Circos is not found.'
else:
msg = 'File etc/housekeeping.conf is not found.'
logger.warning(msg + ' You will have to manually edit etc/housekeeping.conf: '
'set max_points_per_track to ' + str(max_points) + ' and max_ideograms to ' + str(max_ideograms))
return '<<include %s>>\n' % join('etc', 'housekeeping.conf')
housekeeping_fpath = join(output_dir, 'housekeeping.conf')
with open(template_fpath) as f:
with open(housekeeping_fpath, 'w') as out_f:
for line in f:
if 'max_points_per_track' in line:
out_f.write('max_points_per_track = %d\n' % max_points)
elif 'max_ideograms' in line:
out_f.write('max_ideograms = %d\n' % max_ideograms)
else:
out_f.write(line)
return '<<include %s>>\n' % relpath(housekeeping_fpath, root_dir)
def search_sv_with_gridss(main_ref_fpath, bam_fpath, meta_ref_fpaths, output_dirpath, err_fpath):
logger.info(' Searching structural variations with GRIDSS...')
final_bed_fpath = join(output_dirpath, qutils.name_from_fpath(main_ref_fpath) + '_' + qconfig.sv_bed_fname)
if isfile(final_bed_fpath):
logger.info(' Using existing file: ' + final_bed_fpath)
return final_bed_fpath
if not get_path_to_program('java') or not check_java_version(1.8):
logger.warning('Java 1.8 or later is required to run GRIDSS. Please install it and rerun QUAST.')
return None
if not get_path_to_program('Rscript'):
logger.warning('R is required to run GRIDSS. Please install it and rerun QUAST.')
return None
if meta_ref_fpaths:
n_jobs = min(len(meta_ref_fpaths), qconfig.max_threads)
threads_per_job = max(1, qconfig.max_threads // n_jobs)
parallel_args = [(cur_ref_fpath, output_dirpath, err_fpath, threads_per_job) for cur_ref_fpath in meta_ref_fpaths]
bed_fpaths = run_parallel(process_one_ref, parallel_args, n_jobs, filter_results=True)
if bed_fpaths:
qutils.cat_files(bed_fpaths, final_bed_fpath)
else:
process_one_ref(main_ref_fpath, output_dirpath, err_fpath, qconfig.max_threads, bam_fpath=bam_fpath, bed_fpath=final_bed_fpath)
logger.info(' Saving to: ' + final_bed_fpath)
return final_bed_fpath
def compile_glimmer(logger, only_clean=False):
tool_dirpath = os.path.join(qconfig.LIBS_LOCATION, 'glimmer')
tool_src_dirpath = os.path.join(tool_dirpath, 'src')
if not get_path_to_program('glimmerhmm', tool_dirpath):
compile_tool('GlimmerHMM', tool_src_dirpath, ['../glimmerhmm'], logger=logger, only_clean=only_clean)
return get_path_to_program('glimmerhmm', tool_dirpath)
def search_sv_with_gridss(main_ref_fpath, bam_fpath, meta_ref_fpaths, output_dirpath, err_fpath):
logger.info(' Searching structural variations with GRIDSS...')
final_bed_fpath = join(output_dirpath, qutils.name_from_fpath(main_ref_fpath) + '_' + qconfig.sv_bed_fname)
if isfile(final_bed_fpath):
logger.info(' Using existing file: ' + final_bed_fpath)
return final_bed_fpath
if not get_path_to_program('java') or not check_java_version(1.8):
logger.warning('Java 1.8 or later is required to run GRIDSS. Please install it and rerun QUAST.')
return None
if not get_path_to_program('Rscript'):
logger.warning('R is required to run GRIDSS. Please install it and rerun QUAST.')
return None
if meta_ref_fpaths:
n_jobs = min(len(meta_ref_fpaths), qconfig.max_threads)
threads_per_job = max(1, qconfig.max_threads // n_jobs)
parallel_args = [(cur_ref_fpath, output_dirpath, err_fpath, threads_per_job) for cur_ref_fpath in meta_ref_fpaths]
bed_fpaths = run_parallel(process_one_ref, parallel_args, n_jobs, filter_results=True)
if bed_fpaths:
qutils.cat_files(bed_fpaths, final_bed_fpath)
else:
process_one_ref(main_ref_fpath, output_dirpath, err_fpath, qconfig.max_threads, bam_fpath=bam_fpath, bed_fpath=final_bed_fpath)
logger.info(' Saving to: ' + final_bed_fpath)
return final_bed_fpath
def do(ref_fpath, contigs_fpaths, contig_report_fpath_pattern, gc_fpath,
features_containers, cov_fpath, output_dir, logger):
if not exists(output_dir):
os.makedirs(output_dir)
conf_fpath, circos_legend_fpath = create_conf(ref_fpath, contigs_fpaths,
contig_report_fpath_pattern,
output_dir, gc_fpath,
features_containers,
cov_fpath, logger)
circos_exec = get_path_to_program('circos')
if not circos_exec:
logger.warning(
'Circos is not installed!\n'
'If you want to create Circos plots, install Circos as described at http://circos.ca/tutorials/lessons/configuration/distribution_and_installation '
'and run the following command:\n\tcircos -conf ' + conf_fpath +
'\n'
'The plot legend is saved to ' + circos_legend_fpath + '\n')
return None, None
cmdline = [circos_exec, '-conf', conf_fpath]
log_fpath = join(output_dir, 'circos.log')
err_fpath = join(output_dir, 'circos.err')
circos_png_fpath = join(output_dir, circos_png_fname)
return_code = qutils.call_subprocess(cmdline,
stdout=open(log_fpath, 'w'),
stderr=open(err_fpath, 'w'))
if return_code == 0 and is_non_empty_file(circos_png_fpath):
return circos_png_fpath, circos_legend_fpath
else:
logger.warning(' Circos diagram was not created. See ' + log_fpath +
' and ' + err_fpath + ' for details')
return None, None
def get_blast_fpath(fname):
blast_path = os.path.join(blast_dirpath, fname)
if os.path.exists(blast_path):
return blast_path
blast_path = qutils.get_path_to_program(fname)
return blast_path
def get_blast_fpath(fname):
blast_path = os.path.join(blast_dirpath, fname)
if os.path.exists(blast_path):
return blast_path
blast_path = qutils.get_path_to_program(fname)
return blast_path
def gnuplot_exec_fpath():
system_gnuplot_fpath = get_path_to_program('gnuplot')
if system_gnuplot_fpath:
return system_gnuplot_fpath, True
tool_dirpath = join(qconfig.LIBS_LOCATION, 'gnuplot')
tool_src_dirpath = join(tool_dirpath, 'src')
tool_exec_fpath = join(tool_src_dirpath, 'gnuplot')
return tool_exec_fpath, False
def do(ref_fpath, contigs_fpaths, contig_report_fpath_pattern, gc_fpath, features_containers, cov_fpath, output_dir, logger):
if not exists(output_dir):
os.makedirs(output_dir)
conf_fpath, circos_legend_fpath = create_conf(ref_fpath, contigs_fpaths, contig_report_fpath_pattern, output_dir, gc_fpath, features_containers, cov_fpath, logger)
circos_exec = get_path_to_program('circos')
if not circos_exec:
logger.warning('Circos is not installed!\n'
'If you want to create Circos plots, install Circos as described at http://circos.ca/tutorials/lessons/configuration/distribution_and_installation '
'and run the following command:\n circos -conf ' + conf_fpath + '.\n '
'The plot annotation is saved to ' + circos_legend_fpath)
return None, None
cmdline = [circos_exec, '-conf', conf_fpath]
log_fpath = join(output_dir, 'circos.log')
err_fpath = join(output_dir, 'circos.err')
circos_png_fpath = join(output_dir, circos_png_fname)
return_code = qutils.call_subprocess(cmdline, stdout=open(log_fpath, 'w'), stderr=open(err_fpath, 'w'))
if return_code == 0 and is_non_empty_file(circos_png_fpath):
return circos_png_fpath, circos_legend_fpath
else:
logger.warning(' Circos diagram was not created. See ' + log_fpath + ' and ' + err_fpath + ' for details')
return None, None
def download_augustus(logger, only_clean=False):
def __check_preinstalled_augustus_completeness(dirpath):
etraining_path = os.path.join(dirpath, 'bin', 'etraining')
if not os.path.isfile(etraining_path) or not os.access(
etraining_path, os.X_OK):
return False
for aux_dir in ['scripts', 'config']:
aux_dir_path = os.path.join(dirpath, aux_dir)
if not os.path.isdir(aux_dir_path):
return False
return True
preinstalled_augustus = qutils.get_path_to_program('augustus')
if preinstalled_augustus is not None:
preinstalled_augustus_dirpath = os.path.dirname(
os.path.dirname(preinstalled_augustus))
if __check_preinstalled_augustus_completeness(
preinstalled_augustus_dirpath):
return preinstalled_augustus_dirpath
return download_tool('augustus',
augustus_version, ['bin'],
logger,
augustus_url,
only_clean=only_clean)
def get_installed_emem():
return qutils.get_path_to_program('e-mem')
def minimap_fpath():
return get_path_to_program('minimap2', minimap_dirpath)
def bedtools_fpath(fname):
return get_path_to_program(fname, bedtools_bin_dirpath)
def bwa_fpath(fname):
return get_path_to_program(fname, bwa_dirpath)
def get_installed_emem():
return qutils.get_path_to_program('e-mem')
def minimap_fpath():
return get_path_to_program('minimap2',
contig_aligner_dirpath,
min_version='2.10')
def bedtools_fpath(fname):
return get_path_to_program(fname, bedtools_bin_dirpath)
def bwa_fpath(fname):
return get_path_to_program(fname, bwa_dirpath)
def minimap_fpath():
return get_path_to_program('minimap2', contig_aligner_dirpath, min_version='2.10')