Merge pull request #163 from nekrut/bcf_metadata

[WIP] Added default indexing for bcf files

lib/galaxy/datatypes/binary.py

@@ -423,11 +423,12 @@ class Bam( Binary ):
 
 Binary.register_sniffable_binary_format("bam", "bam", Bam)
 
-
 class Bcf( Binary):
     """Class describing a BCF file"""
     file_ext = "bcf"
 
+    MetadataElement( name="bcf_index", desc="BCF Index File", param=metadata.FileParameter, file_ext="csi", readonly=True, no_value=None, visible=False, optional=True )
+
     def sniff( self, filename ):
         # BCF is compressed in the BGZF format, and must not be uncompressed in Galaxy.
         # The first 3 bytes of any bcf file is 'BCF', and the file is binary.
@@ -439,6 +440,37 @@ class Bcf( Binary):
         except:
             return False
 
+    def set_meta( self, dataset, overwrite = True, **kwd ):
+        """ Creates the index for the BCF file. """
+        # These metadata values are not accessible by users, always overwrite
+        index_file = dataset.metadata.bcf_index
+        if not index_file:
+            index_file = dataset.metadata.spec['bcf_index'].param.new_file( dataset = dataset )
+        # Create the bcf index
+        ##$ bcftools index
+        ##Usage: bcftools index <in.bcf>
+
+        dataset_symlink = os.path.join( os.path.dirname( index_file.file_name ),
+                    '__dataset_%d_%s' % ( dataset.id, os.path.basename( index_file.file_name ) ) )
+        os.symlink( dataset.file_name, dataset_symlink )
+           
+        stderr_name = tempfile.NamedTemporaryFile( prefix = "bcf_index_stderr" ).name
+        command = [ 'bcftools', 'index', dataset_symlink ]
+        proc = subprocess.Popen( args=command, stderr=open( stderr_name, 'wb' ) )
+        exit_code = proc.wait()
+        shutil.move( dataset_symlink + '.csi', index_file.file_name )
+
+        stderr = open( stderr_name ).read().strip()
+        if stderr:
+            if exit_code != 0:
+                os.unlink( stderr_name ) #clean up
+                raise Exception, "Error Setting BCF Metadata: %s" % stderr
+            else:
+                print stderr
+        dataset.metadata.bcf_index = index_file
+        # Remove temp file
+        os.unlink( stderr_name )
+
 Binary.register_sniffable_binary_format("bcf", "bcf", Bcf)
 
 

test-data/bcf_index_metadata_test.txt

@@ -0,0 +1 @@
+gzip compressed data, extra field

test/functional/tools/metadata_bcf.xml

@@ -0,0 +1,16 @@
+<tool id="metadata_bcf" name="metadata_BCF" version="1.0.0">
+    <command>file "${input_bcf.metadata.bcf_index}" &gt; "${output_of_input_metadata}"</command>
+    <inputs>
+      <param name="input_bcf" type="data" format="bcf" label="BCF File"/>
+    </inputs>
+    <outputs>
+      <data format="txt" name="output_of_input_metadata" />
+    </outputs>
+    <tests>
+      <test>
+        <param name="input_bcf" value="bcf_index_metadata_test.bcf" ftype="bcf" />
+        <!-- Tests whether the .bcf.csi file is of "gzip compressed data, extra field" type -->
+        <output name="output_of_input_metadata" ftype="txt" file="bcf_index_metadata_test.txt" compare="contains"/>
+      </test>
+    </tests>
+</tool>

test/functional/tools/sample_datatypes_conf.xml

@@ -12,5 +12,6 @@
     <datatype extension="fastqillumina" type="galaxy.datatypes.sequence:FastqIllumina" display_in_upload="true" />
     <datatype extension="sam" type="galaxy.datatypes.tabular:Sam" display_in_upload="true" />
     <datatype extension="bam" type="galaxy.datatypes.binary:Bam" mimetype="application/octet-stream" display_in_upload="true" description="A binary file compressed in the BGZF format with a '.bam' file extension." description_url="https://wiki.galaxyproject.org/Learn/Datatypes#BAM" />
+    <datatype extension="bcf" type="galaxy.datatypes.binary:Bcf" mimetype="application/octet-stream" display_in_upload="true" description="A binary file compressed in the BGZF format with a '.bcf' file extension." description_url="https://wiki.galaxyproject.org/Learn/Datatypes#BCF" />
   </registration>
 </datatypes>

test/functional/tools/samples_tool_conf.xml

@@ -16,6 +16,7 @@
   <tool file="composite_output.xml" />
   <tool file="metadata.xml" />
   <tool file="metadata_bam.xml" />
+  <tool file="metadata_bcf.xml" />
   <tool file="detect_errors_aggressive.xml" />
   <tool file="md5sum.xml" />
   <tool file="job_properties.xml" />