view blast.py @ 11:01b38f20197e draft default tip

v0.0.21 - Updated citation information with GigaScience paper
author peterjc
date Fri, 04 Sep 2015 07:10:04 -0400
parents 5482a8cd0f36
children
line wrap: on
line source

"""
BlastXml class
"""

from galaxy.datatypes.data import get_file_peek
from galaxy.datatypes.data import Text, Data, GenericAsn1
from galaxy.datatypes.xml import GenericXml
from galaxy.datatypes.metadata import MetadataElement

from time import sleep
import os
import logging

log = logging.getLogger(__name__)

class BlastXml( GenericXml ):
    """NCBI Blast XML Output data"""
    file_ext = "blastxml"

    def set_peek( self, dataset, is_multi_byte=False ):
        """Set the peek and blurb text"""
        if not dataset.dataset.purged:
            dataset.peek = get_file_peek( dataset.file_name, is_multi_byte=is_multi_byte )
            dataset.blurb = 'NCBI Blast XML data'
        else:
            dataset.peek = 'file does not exist'
            dataset.blurb = 'file purged from disk'

    def sniff( self, filename ):
        """
        Determines whether the file is blastxml
        
        >>> fname = get_test_fname( 'megablast_xml_parser_test1.blastxml' )
        >>> BlastXml().sniff( fname )
        True
        >>> fname = get_test_fname( 'tblastn_four_human_vs_rhodopsin.xml' )
        >>> BlastXml().sniff( fname )
        True
        >>> fname = get_test_fname( 'interval.interval' )
        >>> BlastXml().sniff( fname )
        False
        """
        #TODO - Use a context manager on Python 2.5+ to close handle
        handle = open(filename)
        line = handle.readline()
        if line.strip() != '<?xml version="1.0"?>':
            handle.close()
            return False
        line = handle.readline()
        if line.strip() not in ['<!DOCTYPE BlastOutput PUBLIC "-//NCBI//NCBI BlastOutput/EN" "http://www.ncbi.nlm.nih.gov/dtd/NCBI_BlastOutput.dtd">',
                                '<!DOCTYPE BlastOutput PUBLIC "-//NCBI//NCBI BlastOutput/EN" "NCBI_BlastOutput.dtd">']:
            handle.close()
            return False
        line = handle.readline()
        if line.strip() != '<BlastOutput>':
            handle.close()
            return False
        handle.close()
        return True
    
    def merge(split_files, output_file):
        """Merging multiple XML files is non-trivial and must be done in subclasses."""
        if len(split_files) == 1:
            #For one file only, use base class method (move/copy)
            return Text.merge(split_files, output_file)
        if not split_files:
            raise ValueError("Given no BLAST XML files, %r, to merge into %s" \
                             % (split_files, output_file))
        out = open(output_file, "w")
        h = None
        for f in split_files:
            if not os.path.isfile(f):
                log.warning("BLAST XML file %s missing, retry in 1s..." % f)
                sleep(1)
            if not os.path.isfile(f):
                log.error("BLAST XML file %s missing" % f)
                raise ValueError("BLAST XML file %s missing" % f)
            h = open(f)
            header = h.readline()
            if not header:
                out.close()
                h.close()
                #Retry, could be transient error with networked file system...
                log.warning("BLAST XML file %s empty, retry in 1s..." % f)
                sleep(1)
                h = open(f)
                header = h.readline()
                if not header:
                    log.error("BLAST XML file %s was empty" % f)
                    raise ValueError("BLAST XML file %s was empty" % f)
            if header.strip() != '<?xml version="1.0"?>':
                out.write(header) #for diagnosis
                out.close()
                h.close()
                raise ValueError("%s is not an XML file!" % f)
            line = h.readline()
            header += line
            if line.strip() not in ['<!DOCTYPE BlastOutput PUBLIC "-//NCBI//NCBI BlastOutput/EN" "http://www.ncbi.nlm.nih.gov/dtd/NCBI_BlastOutput.dtd">',
                                    '<!DOCTYPE BlastOutput PUBLIC "-//NCBI//NCBI BlastOutput/EN" "NCBI_BlastOutput.dtd">']:
                out.write(header) #for diagnosis
                out.close()
                h.close()
                raise ValueError("%s is not a BLAST XML file!" % f)
            while True:
                line = h.readline()
                if not line:
                    out.write(header) #for diagnosis
                    out.close()
                    h.close()
                    raise ValueError("BLAST XML file %s ended prematurely" % f)
                header += line
                if "<Iteration>" in line:
                    break
                if len(header) > 10000:
                    #Something has gone wrong, don't load too much into memory!
                    #Write what we have to the merged file for diagnostics
                    out.write(header)
                    out.close()
                    h.close()
                    raise ValueError("BLAST XML file %s has too long a header!" % f)
            if "<BlastOutput>" not in header:
                out.close()
                h.close()
                raise ValueError("%s is not a BLAST XML file:\n%s\n..." % (f, header))
            if f == split_files[0]:
                out.write(header)
                old_header = header
            elif old_header[:300] != header[:300]:
                #Enough to check <BlastOutput_program> and <BlastOutput_version> match
                out.close()
                h.close()
                raise ValueError("BLAST XML headers don't match for %s and %s - have:\n%s\n...\n\nAnd:\n%s\n...\n" \
                                 % (split_files[0], f, old_header[:300], header[:300]))
            else:
                out.write("    <Iteration>\n")
            for line in h:
                if "</BlastOutput_iterations>" in line:
                    break
                #TODO - Increment <Iteration_iter-num> and if required automatic query names
                #like <Iteration_query-ID>Query_3</Iteration_query-ID> to be increasing?
                out.write(line)
            h.close()
        out.write("  </BlastOutput_iterations>\n")
        out.write("</BlastOutput>\n")
        out.close()
    merge = staticmethod(merge)


class _BlastDb(object):
    """Base class for BLAST database datatype."""

    def set_peek( self, dataset, is_multi_byte=False ):
        """Set the peek and blurb text."""
        if not dataset.dataset.purged:
            dataset.peek  = "BLAST database (multiple files)"
            dataset.blurb = "BLAST database (multiple files)"
        else:
            dataset.peek = 'file does not exist'
            dataset.blurb = 'file purged from disk'

    def display_peek( self, dataset ):
        """Create HTML content, used for displaying peek."""
        try:
            return dataset.peek
        except:
            return "BLAST database (multiple files)"

    def display_data(self, trans, data, preview=False, filename=None,
                     to_ext=None, size=None, offset=None, **kwd):
        """Documented as an old display method, but still gets called via tests etc

        This allows us to format the data shown in the central pane via the "eye" icon.
        """
        if filename is not None and filename != "index":
            #Change nothing - important for the unit tests to access child files:
            return Data.display_data(self, trans, data, preview, filename,
                                     to_ext, size, offset, **kwd)
        if self.file_ext == "blastdbn":
            title = "This is a nucleotide BLAST database"
        elif self.file_ext =="blastdbp":
            title = "This is a protein BLAST database"
        elif self.file_ext =="blastdbd":
            title = "This is a domain BLAST database"
        else:
            #Error?
            title = "This is a BLAST database."
        msg = ""
        try:
            #Try to use any text recorded in the dummy index file:
            handle = open(data.file_name, "rU")
            msg = handle.read().strip()
            handle.close()
        except Exception, err:
            #msg = str(err)
            pass
        if not msg:
            msg = title
        #Galaxy assumes HTML for the display of composite datatypes,
        return "<html><head><title>%s</title></head><body><pre>%s</pre></body></html>" % (title, msg)

    def merge(split_files, output_file):
        """Merge BLAST databases (not implemented for now)."""
        raise NotImplementedError("Merging BLAST databases is non-trivial (do this via makeblastdb?)")

    def split( cls, input_datasets, subdir_generator_function, split_params):
        """Split a BLAST database (not implemented for now)."""
        if split_params is None:
            return None
        raise NotImplementedError("Can't split BLAST databases")


class BlastNucDb( _BlastDb, Data ):
    """Class for nucleotide BLAST database files."""
    file_ext = 'blastdbn'
    allow_datatype_change = False
    composite_type = 'basic'

    def __init__(self, **kwd):
        Data.__init__(self, **kwd)
        self.add_composite_file('blastdb.nhr', is_binary=True) # sequence headers
        self.add_composite_file('blastdb.nin', is_binary=True) # index file
        self.add_composite_file('blastdb.nsq', is_binary=True) # nucleotide sequences
        self.add_composite_file('blastdb.nal', is_binary=False, optional=True) # alias ( -gi_mask option of makeblastdb)
        self.add_composite_file('blastdb.nhd', is_binary=True, optional=True) # sorted sequence hash values ( -hash_index option of makeblastdb)
        self.add_composite_file('blastdb.nhi', is_binary=True, optional=True) # index of sequence hash values ( -hash_index option of makeblastdb)
        self.add_composite_file('blastdb.nnd', is_binary=True, optional=True) # sorted GI values ( -parse_seqids option of makeblastdb and gi present in the description lines)
        self.add_composite_file('blastdb.nni', is_binary=True, optional=True) # index of GI values ( -parse_seqids option of makeblastdb and gi present in the description lines)
        self.add_composite_file('blastdb.nog', is_binary=True, optional=True) # OID->GI lookup file ( -hash_index or -parse_seqids option of makeblastdb)
        self.add_composite_file('blastdb.nsd', is_binary=True, optional=True) # sorted sequence accession values ( -hash_index or -parse_seqids option of makeblastdb)
        self.add_composite_file('blastdb.nsi', is_binary=True, optional=True) # index of sequence accession values ( -hash_index or -parse_seqids option of makeblastdb)
#        self.add_composite_file('blastdb.00.idx', is_binary=True, optional=True) # first volume of the MegaBLAST index generated by makembindex
# The previous line should be repeated for each index volume, with filename extensions like '.01.idx', '.02.idx', etc.
        self.add_composite_file('blastdb.shd', is_binary=True, optional=True) # MegaBLAST index superheader (-old_style_index false option of makembindex)
#        self.add_composite_file('blastdb.naa', is_binary=True, optional=True) # index of a WriteDB column for e.g. mask data
#        self.add_composite_file('blastdb.nab', is_binary=True, optional=True) # data of a WriteDB column
#        self.add_composite_file('blastdb.nac', is_binary=True, optional=True) # multiple byte order for a WriteDB column
# The previous 3 lines should be repeated for each WriteDB column, with filename extensions like ('.nba', '.nbb', '.nbc'), ('.nca', '.ncb', '.ncc'), etc.


class BlastProtDb( _BlastDb, Data ):
    """Class for protein BLAST database files."""
    file_ext = 'blastdbp'
    allow_datatype_change = False
    composite_type = 'basic'

    def __init__(self, **kwd):
        Data.__init__(self, **kwd)
# Component file comments are as in BlastNucDb except where noted
        self.add_composite_file('blastdb.phr', is_binary=True)
        self.add_composite_file('blastdb.pin', is_binary=True)
        self.add_composite_file('blastdb.psq', is_binary=True) # protein sequences
        self.add_composite_file('blastdb.phd', is_binary=True, optional=True)
        self.add_composite_file('blastdb.phi', is_binary=True, optional=True)
        self.add_composite_file('blastdb.pnd', is_binary=True, optional=True)
        self.add_composite_file('blastdb.pni', is_binary=True, optional=True)
        self.add_composite_file('blastdb.pog', is_binary=True, optional=True)
        self.add_composite_file('blastdb.psd', is_binary=True, optional=True)
        self.add_composite_file('blastdb.psi', is_binary=True, optional=True)
#        self.add_composite_file('blastdb.paa', is_binary=True, optional=True)
#        self.add_composite_file('blastdb.pab', is_binary=True, optional=True)
#        self.add_composite_file('blastdb.pac', is_binary=True, optional=True)
# The last 3 lines should be repeated for each WriteDB column, with filename extensions like ('.pba', '.pbb', '.pbc'), ('.pca', '.pcb', '.pcc'), etc.


class BlastDomainDb( _BlastDb, Data ):
    """Class for domain BLAST database files."""
    file_ext = 'blastdbd'
    allow_datatype_change = False
    composite_type = 'basic'

    def __init__(self, **kwd):
        Data.__init__(self, **kwd)
        self.add_composite_file('blastdb.phr', is_binary=True)
        self.add_composite_file('blastdb.pin', is_binary=True)
        self.add_composite_file('blastdb.psq', is_binary=True)
        self.add_composite_file('blastdb.freq', is_binary=True, optional=True)
        self.add_composite_file('blastdb.loo', is_binary=True, optional=True)
        self.add_composite_file('blastdb.psd', is_binary=True, optional=True)
        self.add_composite_file('blastdb.psi', is_binary=True, optional=True)
        self.add_composite_file('blastdb.rps', is_binary=True, optional=True)
        self.add_composite_file('blastdb.aux', is_binary=True, optional=True)