Merge pull request #7299 from chambm/feature/rawtar-datatypes

Add datatypes and sniffers for tar archives of directory-based MS formats and WIFF/SCAN pairs

config/datatypes_conf.xml.sample

@@ -219,6 +219,12 @@
     <datatype extension="idxml" type="galaxy.datatypes.proteomics:IdXML" mimetype="application/xml" display_in_upload="true"/>
     <datatype extension="tandem" type="galaxy.datatypes.proteomics:TandemXML" mimetype="application/xml" display_in_upload="true"/>
     <datatype extension="thermo.raw" type="galaxy.datatypes.proteomics:ThermoRAW" mimetype="application/octet-stream" display_in_upload="true"/>
+    <datatype extension="brukerbaf.d.tar" type="galaxy.datatypes.binary:BafTar" display_in_upload="true"/>
+    <datatype extension="agilentbrukeryep.d.tar" type="galaxy.datatypes.binary:YepTar" display_in_upload="true"/>
+    <datatype extension="brukertdf.d.tar" type="galaxy.datatypes.binary:TdfTar" display_in_upload="true"/>
+    <datatype extension="agilentmasshunter.d.tar" type="galaxy.datatypes.binary:MassHunterTar" display_in_upload="true"/>
+    <datatype extension="watersmasslynx.raw.tar" type="galaxy.datatypes.binary:MassLynxTar" display_in_upload="true"/>
+    <datatype extension="wiff.tar" type="galaxy.datatypes.binary:WiffTar" display_in_upload="true"/>
     <datatype extension="mzml" type="galaxy.datatypes.proteomics:MzML" mimetype="application/xml" display_in_upload="true"/>
     <datatype extension="nmrml" type="galaxy.datatypes.proteomics:NmrML" mimetype="application/xml" display_in_upload="true" description="nmrML is an open mark-up language for NMR data." description_url="http://nmrml.org/schema/"/>
     <datatype extension="mgf" type="galaxy.datatypes.proteomics:Mgf" display_in_upload="true"/>
@@ -745,6 +751,12 @@
     <sniffer type="galaxy.datatypes.binary:DAA"/>
     <sniffer type="galaxy.datatypes.binary:RMA6"/>
     <sniffer type="galaxy.datatypes.binary:DMND"/>
+    <sniffer type="galaxy.datatypes.binary:BafTar"/>
+    <sniffer type="galaxy.datatypes.binary:TdfTar"/>
+    <sniffer type="galaxy.datatypes.binary:MassHunterTar"/>
+    <sniffer type="galaxy.datatypes.binary:MassLynxTar"/>
+    <sniffer type="galaxy.datatypes.binary:YepTar"/>
+    <sniffer type="galaxy.datatypes.binary:WiffTar"/>
     <sniffer type="galaxy.datatypes.binary:Fast5ArchiveGz"/>
     <sniffer type="galaxy.datatypes.binary:Fast5ArchiveBz2"/>
     <sniffer type="galaxy.datatypes.binary:Fast5Archive"/>

lib/galaxy/datatypes/binary.py

@@ -2239,6 +2239,118 @@ def sniff(self, dataset):
         return False
 
 
+class BafTar(CompressedArchive):
+    """
+    Base class for common behavior of tar files of directory-based raw file formats
+    >>> from galaxy.datatypes.sniff import get_test_fname
+    >>> fname = get_test_fname('brukerbaf.d.tar')
+    >>> BafTar().sniff(fname)
+    True
+    >>> fname = get_test_fname('test.fast5.tar')
+    >>> BafTar().sniff(fname)
+    False
+    """
+    edam_data = "data_2536"  # mass spectrometry data
+    edam_format = "format_3712"  # TODO: add more raw formats to EDAM?
+    file_ext = "brukerbaf.d.tar"
+
+    def get_signature_file(self):
+        return "analysis.baf"
+
+    def sniff(self, filename):
+        if tarfile.is_tarfile(filename):
+            with tarfile.open(filename) as rawtar:
+                return self.get_signature_file() in [os.path.basename(f).lower() for f in rawtar.getnames()]
+        return False
+
+    def get_type(self):
+        return "Bruker BAF directory archive"
+
+    def set_peek(self, dataset, is_multi_byte=False):
+        if not dataset.dataset.purged:
+            dataset.peek = self.get_type()
+            dataset.blurb = nice_size(dataset.get_size())
+        else:
+            dataset.peek = 'file does not exist'
+            dataset.blurb = 'file purged from disk'
+
+    def display_peek(self, dataset):
+        try:
+            return dataset.peek
+        except Exception:
+            return "%s (%s)" % (self.get_type(), nice_size(dataset.get_size()))
+
+
+class YepTar(BafTar):
+    """ A tar'd up .d directory containing Agilent/Bruker YEP format data """
+    file_ext = "agilentbrukeryep.d.tar"
+
+    def get_signature_file(self):
+        return "analysis.yep"
+
+    def get_type(self):
+        return "Agilent/Bruker YEP directory archive"
+
+
+class TdfTar(BafTar):
+    """ A tar'd up .d directory containing Bruker TDF format data """
+    file_ext = "brukertdf.d.tar"
+
+    def get_signature_file(self):
+        return "analysis.tdf"
+
+    def get_type(self):
+        return "Bruker TDF directory archive"
+
+
+class MassHunterTar(BafTar):
+    """ A tar'd up .d directory containing Agilent MassHunter format data """
+    file_ext = "agilentmasshunter.d.tar"
+
+    def get_signature_file(self):
+        return "msscan.bin"
+
+    def get_type(self):
+        return "Agilent MassHunter directory archive"
+
+
+class MassLynxTar(BafTar):
+    """ A tar'd up .d directory containing Waters MassLynx format data """
+    file_ext = "watersmasslynx.raw.tar"
+
+    def get_signature_file(self):
+        return "_func001.dat"
+
+    def get_type(self):
+        return "Waters MassLynx RAW directory archive"
+
+
+class WiffTar(BafTar):
+    """
+    A tar'd up .wiff/.scan pair containing Sciex WIFF format data
+    >>> from galaxy.datatypes.sniff import get_test_fname
+    >>> fname = get_test_fname('some.wiff.tar')
+    >>> WiffTar().sniff(fname)
+    True
+    >>> fname = get_test_fname('brukerbaf.d.tar')
+    >>> WiffTar().sniff(fname)
+    False
+    >>> fname = get_test_fname('test.fast5.tar')
+    >>> WiffTar().sniff(fname)
+    False
+    """
+    file_ext = "wiff.tar"
+
+    def sniff(self, filename):
+        if tarfile.is_tarfile(filename):
+            with tarfile.open(filename) as rawtar:
+                return ".wiff" in [os.path.splitext(os.path.basename(f).lower())[1] for f in rawtar.getnames()]
+        return False
+
+    def get_type(self):
+        return "Sciex WIFF/SCAN archive"
+
+
 if __name__ == '__main__':
     import doctest
     doctest.testmod(sys.modules[__name__])