closes #11823

[arvados.git] / crunch_scripts / arvados_samtools.py

diff --git a/crunch_scripts/arvados_samtools.py b/crunch_scripts/arvados_samtools.py
index dd7a42874f9b3dbe4863e2bc752fda09573fb1e0..6f4d96645839049dae8ab35366c465e974aeab65 100644 (file)
--- a/crunch_scripts/arvados_samtools.py
+++ b/crunch_scripts/arvados_samtools.py
@@ -8,6 +8,10 @@ import subprocess
 samtools_path = None
 
 def samtools_install_path():
+    """
+    Extract the samtools source tree, build the samtools binary, and
+    return the path to the source tree.
+    """
     global samtools_path
     if samtools_path:
         return samtools_path
@@ -25,9 +29,24 @@ def samtools_install_path():
     return samtools_path
 
 def samtools_binary():
+    """
+    Return the path to the samtools executable.
+    """
     return os.path.join(samtools_install_path(), 'samtools')
 
 def run(command, command_args, **kwargs):
+    """
+    Build and run the samtools binary.
+
+    command is the samtools subcommand, e.g., "view" or "sort".
+
+    command_args is a list of additional command line arguments, e.g.,
+    ['-bt', 'ref_list.txt', '-o', 'aln.bam', 'aln.sam.gz']
+
+    It is assumed that we are running in a Crunch job environment, and
+    the job's "samtools_tgz" parameter is a collection containing the
+    samtools source tree in a .tgz file.
+    """
     execargs = [samtools_binary(),
                 command]
     execargs += command_args
@@ -40,6 +59,20 @@ def run(command, command_args, **kwargs):
         stdout=kwargs.get('stdout', sys.stderr))
 
 def one_task_per_bam_file(if_sequence=0, and_end_task=True):
+    """
+    Queue one task for each bam file in this job's input collection.
+
+    Each new task will have an "input" parameter: a manifest
+    containing one .bam file and (if available) the corresponding .bai
+    index file.
+
+    Files in the input collection that are not named *.bam or *.bai
+    (as well as *.bai files that do not match any .bam file present)
+    are silently ignored.
+
+    if_sequence and and_end_task arguments have the same significance
+    as in arvados.job_setup.one_task_per_input_file().
+    """
     if if_sequence != arvados.current_task()['sequence']:
         return
     job_input = arvados.current_job()['script_parameters']['input']