Retrieve the SRA public dir from a configuration

Author: arteymix

example.luigi.cfg

@@ -54,6 +54,11 @@ RSEM_DIR=contrib/RSEM
 
 SLACK_WEBHOOK_URL=
 
+[rnaseq_pipeline.sources.sra]
+# location where tools like prefetch and fastq-dump will store downloaded SRA files
+# you can get this value with vdb-config -p
+ncbi_public_dir=/cosmos/scratch/ncbi/public
+
 [rnaseq_pipeline.gemma]
 cli_bin=gemma-cli
 # values for $JAVA_HOME and $JAVA_OPTS environment variables

rnaseq_pipeline/sources/sra.py

@@ -28,6 +28,12 @@
 
 logger = logging.getLogger(__name__)
 
+class sra(luigi.Config):
+    task_namespace = 'rnaseq_pipeline.sources'
+    ncbi_public_dir: str = luigi.Parameter()
+
+sra_config = sra()
+
 # columns to use when a runinfo file lacks a header
 SRA_RUNINFO_COLUMNS = ['Run', 'ReleaseDate', 'LoadDate', 'spots', 'bases', 'spots_with_mates', 'avgLength', 'size_MB',
                        'AssemblyName', 'download_path', 'Experiment', 'LibraryName', 'LibraryStrategy',
@@ -164,12 +170,6 @@ class PrefetchSraRun(TaskWithMetadataMixin, luigi.Task):
 
     retry_count = 3
 
-    @staticmethod
-    def _get_ncbi_public_dir():
-        ret = subprocess.run(['vdb-config', '-p'], stdout=subprocess.PIPE, universal_newlines=True)
-        config_xml = ET.fromstring(ret.stdout)
-        return config_xml.find('repository').find('user').find('main').find('public').find('root').text
-
     def run(self):
         yield sratoolkit.Prefetch(srr_accession=self.srr,
                                   output_file=self.output().path,
@@ -179,7 +179,7 @@ def run(self):
                                   walltime=timedelta(hours=2))
 
     def output(self):
-        return luigi.LocalTarget(join(self._get_ncbi_public_dir(), 'sra', f'{self.srr}.sra'))
+        return luigi.LocalTarget(join(sra_config.ncbi_public_dir, 'sra', f'{self.srr}.sra'))
 
 @requires(PrefetchSraRun)
 class DumpSraRun(luigi.Task):