Merge pull request #13 from Clinical-Genomics/export

Automate the export of datasets

mutacc_auto/cli/export_command.py

@@ -0,0 +1,93 @@
+import click
+import logging
+from pathlib import Path
+
+import yaml
+
+from mutacc_auto.recipes.export_recipe import export_dataset
+from mutacc_auto.utils.tmp_dir import TemporaryDirectory
+
+MUTACC_TMP = 'temporaries'
+MUTACC_ROOT_DIR = 'root_dir'
+
+LOG = logging.getLogger(__name__)
+
+def parse_path(ctx, param, value):
+
+    if value:
+        value = str(Path(str(value)).expanduser().absolute().resolve())
+    return value
+
+@click.command('export')
+@click.option('-c', '--mutacc-config',
+              type=click.Path(exists=True),
+              callback=parse_path,
+              help="configuration file used for mutacc")
+@click.option('-o', '--vcf-out',
+              type=click.Path(exists=False),
+              callback=parse_path,
+              help="Path to created vcf-file")
+@click.option('-b', '--background',
+              type=click.Path(exists=True),
+              help="yaml file with genomic backgrounds for each sample in trio")
+@click.option('-k', '--conda',
+              is_flag=True,
+              help="Use 'conda activate' to source environment")
+@click.option('-e','--environment',
+              type=str,
+              help="conda environment used for mutacc")
+@click.option('-D','--dry',
+              is_flag=True,
+              help="dry run")
+@click.option('-V','--verbose',
+              is_flag=True,
+              help="verbose")
+@click.pass_context
+def export_command(ctx, mutacc_config, vcf_out, background, conda, environment, dry, verbose):
+
+    mutacc_config = mutacc_config or ctx.obj['mutacc_config']
+    mutacc_binary = ctx.obj.get('mutacc_binary')
+
+    with open(background, 'r') as background_handle:
+        background_datasets = yaml.load(background_handle)
+
+    slurm_config = {}
+    if ctx.obj.get('slurm'):
+        slurm_config = ctx.obj['slurm']
+
+    slurm_options = {}
+    slurm_options['log_directory'] = slurm_config['log_directory']
+    slurm_options['email'] = slurm_config['email']
+    slurm_options['time'] = slurm_config['time']
+    slurm_options['account'] = slurm_config['account']
+    slurm_options['priority'] = slurm_config['priority']
+
+    with open(Path(mutacc_config)) as yaml_handle:
+        mutacc_config_dict = yaml.load(yaml_handle)
+        mutacc_tmp = Path(mutacc_config_dict[MUTACC_ROOT_DIR]).joinpath(MUTACC_TMP)
+
+    mutacc_auto_tmp_dir = mutacc_tmp.joinpath('mutacc_auto')
+    if not mutacc_auto_tmp_dir.is_dir():
+        mutacc_auto_tmp_dir.mkdir(parents=True)
+
+    with TemporaryDirectory(directory=mutacc_auto_tmp_dir) as tmp_dir:
+
+        LOG.info("All files are placed in {}".format(tmp_dir))
+        sbatch_files = export_dataset(mutacc_config=mutacc_config,
+                                      background=background_datasets,
+                                      mutacc_binary=mutacc_binary,
+                                      case_query='{}',
+                                      merged_vcf_path=vcf_out,
+                                      slurm_options=slurm_options,
+                                      tmp_dir=tmp_dir,
+                                      environment=environment,
+                                      conda=conda,
+                                      dry=dry)
+
+    if verbose:
+
+        for key, value in sbatch_files.items():
+
+            with open(value) as sbatch_handle:
+
+                LOG.info("SBATCH SCRIPT FOR {}, {}\n{}".format(key, value, sbatch_handle.read()))

mutacc_auto/cli/root.py

@@ -7,6 +7,7 @@
 from mutacc_auto import __version__
 from .extract_command import extract_command
 from .import_command import import_command
+from .export_command import export_command
 
 LOG_LEVELS = ['DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL']
 LOG = logging.getLogger(__name__)
@@ -36,3 +37,4 @@ def cli(ctx, loglevel, config_file):
 
 cli.add_command(extract_command)
 cli.add_command(import_command)
+cli.add_command(export_command)

mutacc_auto/commands/command.py

@@ -73,7 +73,7 @@ def check_output(self):
 
         command_stdout = command_stdout.decode('utf-8')
 
-        LOG.debug("Executed:".format(str(self)))
+        LOG.debug("Executed: {}".format(str(self)))
 
         return command_stdout
 
@@ -96,6 +96,6 @@ def call(self):
 
             raise
 
-        LOG.debug("Executed:".format(str(self)))
+        LOG.debug("Executed: {}".format(str(self)))
 
         return result

mutacc_auto/commands/mutacc_command.py

@@ -37,3 +37,41 @@ def __init__(self, config_file, extracted_case_file, mutacc_binary=None):
         self.add_subcommand('db')
         self.add_subcommand('import')
         self.add_argument(str(extracted_case_file))
+
+class MutaccExport(MutaccCommand):
+
+    def __init__(self, config_file, mutacc_binary=None, case_query=None, variant_query=None,
+                 proband=False, member='affected', sample_name=None):
+
+        super(MutaccExport, self).__init__(config_file, mutacc_binary=mutacc_binary)
+
+        self.add_subcommand('db')
+        self.add_subcommand('export')
+
+        if variant_query is not None:
+            self.add_option('variant-query', variant_query)
+        if case_query is not None:
+            self.add_option('case-query', case_query)
+        if sample_name is not None:
+            self.add_option('sample-name', sample_name)
+        if proband:
+            self.add_option('proband')
+
+        self.add_option('member', member)
+        self.add_option('json-out')
+
+class MutaccSynthesize(MutaccCommand):
+
+    def __init__(self, config_file, fastq1, bam_file, query_file, fastq2=None,
+                 mutacc_binary=None):
+
+        super(MutaccSynthesize, self).__init__(config_file, mutacc_binary=mutacc_binary)
+
+        self.add_subcommand('synthesize')
+
+        self.add_option('background-bam', bam_file)
+        self.add_option('background-fastq', fastq1)
+        if fastq2 is not None:
+            self.add_option('background-fastq2', fastq2)
+        self.add_option('query', query_file)
+        self.add_option('json-out')

mutacc_auto/commands/vcf_command.py

@@ -0,0 +1,37 @@
+from .command import Command as BaseCommand
+
+BGZIP_COMMAND = 'bgzip'
+TABIX_COMMAND = 'tabix'
+BCFTOOLS_MERGE_COMMAND = 'bcftools'
+
+class BgzipCommand(BaseCommand):
+
+    def __init__(self, vcf_file):
+
+        super(BgzipCommand, self).__init__(BGZIP_COMMAND)
+
+        self.add_argument(vcf_file)
+
+class TabixCommand(BaseCommand):
+
+    def __init__(self, vcf_file):
+
+        super(TabixCommand, self).__init__(TABIX_COMMAND)
+
+        self.add_argument(vcf_file)
+
+class BcftoolsMergeCommand(BaseCommand):
+
+    def __init__(self, vcf_files, out_vcf=None):
+
+        super(BcftoolsMergeCommand, self).__init__(BCFTOOLS_MERGE_COMMAND)
+
+        self.add_subcommand('merge')
+
+        if out_vcf:
+            self.add_option('output', out_vcf)
+
+        self.add_option('output-type', 'z')
+
+        for vcf_file in vcf_files:
+            self.add_argument(vcf_file)