def countHits(infile, **kwargs):
"""
Count hits from a hit table.
Calls edl.hits.parseM8FileIter with the following optional parameters:
hitStringMap (None): dictionary (or file) mapping hit IDs to
something else
format (GENE): hit table format
filter_top_pct (0): only consider hits within this % of top score for
each read
parseStyle (ACCS): how to process hit data into an identifying string
countMethod ('all'): how to resolve hits to multiple sequences
taxonomy (None): An edl.taxon.Taxonomy object or directory
conatining taxdmp
rank (None): Maximum rank to resolve hits
"""
# if taxonomy or hitStringMap are file names, parse them
taxonomy = kwargs.pop('taxonomy', None)
if isinstance(taxonomy, str):
taxonomy = readTaxonomy(
taxonomy, namesMap=kwargs.pop(
'namesMap', False))
hitStringMap = kwargs.pop('hitStringMap', None)
if isinstance(hitStringMap, str):
if taxonomy is not None:
# the mapped hit ids will need to be ints
valueType = kwargs.pop('valueType', int)
else:
valueType = kwargs.pop('valueType', None)
hitStringMap = parseMapFile(hitStringMap, valueType=valueType)
# if infile is name (and not handle), open as a handle
if isinstance(infile, str):
inhandle = open(infile)
else:
inhandle = infile
# get iterator over reads that will parse hits
hitIter = parseM8FileIter(inhandle,
hitStringMap,
FilterParams(
format=kwargs.pop('format', GENE),
top_pct=kwargs.pop('filter_top_pct', 0),
),
kwargs.pop('parseStyle', ACCS),
kwargs.pop('countMethod', 'all'),
taxonomy=taxonomy,
rank=kwargs.pop('rank', None))
# count the hits
(total, counts) = countIterHits(hitIter,
allMethod=kwargs.pop('allMethod', ALLEQ),
returnMap=False)
logger.info("Total hits: %s" % total)
if isinstance(infile, str):
inhandle.close()
return counts
def countHits(infile, **kwargs):
"""
Count hits from a hit table.
Calls edl.hits.parseM8FileIter with the following optional parameters:
hitStringMap (None): dictionary (or file) mapping hit IDs to
something else
format (GENE): hit table format
filter_top_pct (0): only consider hits within this % of top score for
each read
parseStyle (ACCS): how to process hit data into an identifying string
countMethod ('all'): how to resolve hits to multiple sequences
taxonomy (None): An edl.taxon.Taxonomy object or directory
conatining taxdmp
rank (None): Maximum rank to resolve hits
"""
# if taxonomy or hitStringMap are file names, parse them
taxonomy = kwargs.pop('taxonomy', None)
if isinstance(taxonomy, str):
taxonomy = readTaxonomy(taxonomy,
namesMap=kwargs.pop('namesMap', False))
hitStringMap = kwargs.pop('hitStringMap', None)
if isinstance(hitStringMap, str):
if taxonomy is not None:
# the mapped hit ids will need to be ints
valueType = kwargs.pop('valueType', int)
else:
valueType = kwargs.pop('valueType', None)
hitStringMap = parseMapFile(hitStringMap, valueType=valueType)
# if infile is name (and not handle), open as a handle
if isinstance(infile, str):
inhandle = open(infile)
else:
inhandle = infile
# get iterator over reads that will parse hits
hitIter = parseM8FileIter(inhandle,
hitStringMap,
FilterParams(
format=kwargs.pop('format', GENE),
top_pct=kwargs.pop('filter_top_pct', 0),
),
kwargs.pop('parseStyle', ACCS),
kwargs.pop('countMethod', 'all'),
taxonomy=taxonomy,
rank=kwargs.pop('rank', None))
# count the hits
(total, counts) = countIterHits(hitIter,
allMethod=kwargs.pop('allMethod', ALLEQ),
returnMap=False)
logger.info("Total hits: %s" % total)
if isinstance(infile, str):
inhandle.close()
return counts
def main():
"""" Set up the CLI """
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("input_files", nargs="+",
default=[],
metavar="INFILE",
help="List of hit tables to process")
parser.add_argument("-o", "--outfile", dest="outfile",
metavar="OUTFILE",
help="Write count table to OUTFILE")
parser.add_argument("-r", "--rank", dest="ranks", default=None,
metavar="RANK", action="append",
help=""" Rank(s) to collect counts on. Use flag
multiple
times to specify multiple ranks. If multiple values
given, one table produced for each with rank name
appended to file name. Defaults to all major ranks
between phylum and species. Corresponds to rank names
in nodes.dmp. To see list run:
'cut -f5 nodes.dmp | uniq | sort | uniq'
in ncbi tax dir. Will also accept 'organism' to mean
no rank (ie, just the organism name).""")
parser.add_argument(
"-s",
"--collapseToDomain",
default=False,
action="store_true",
help="Collapse all taxa below given rank down to "
"superkingdom/domain. EG: in the genus output, anything "
"assigned to Cyanobactia, will be lumped in with all "
"other bacteria")
parser.add_argument(
"--proportional",
dest="proportional",
default=False,
action="store_true",
help="""When using tophit or toporg, redistribute proportionally
instead of winner take all""")
parser.add_argument(
"-R",
"--printRank",
dest="printRanks",
action="append",
help="Include indeicated rank(s) in lineage of printed taxa. "
"Will be ignored if beyond the rank of the taxa "
"(IE We can't include species if the taxon being counted "
"is genus)")
# option for deconvoluting clusters or assemblies
add_weight_arguments(parser, multiple=True)
# cutoff options
add_count_arguments(parser)
# format, tax dir, and more
add_taxon_arguments(
parser,
choices={
'countMethod': (
'LCA',
'all',
'first',
'most',
'tophit',
'toporg',
'consensus')})
# log level and help
add_universal_arguments(parser)
arguments = parser.parse_args()
setup_logging(arguments)
if arguments.proportional and \
arguments.countMethod not in ['tophit', 'toporg']:
parser.error("--proportinal only has meaning "
"if using tophit or toporg")
if len(arguments.input_files) == 0:
parser.error("Must supply at least one m8 file to parse")
# Handle the case where Galaxy tries to set None as a string
arguments.ranks = checkNoneOption(arguments.ranks)
arguments.printRanks = checkNoneOption(arguments.printRanks)
logging.info("Printing out ranks: %r", arguments.ranks)
# Set defaults and check for some conflicts
if arguments.ranks is None and arguments.taxdir is None:
# using hit names only
arguments.ranks = [ORG_RANK]
if arguments.printRanks is not None:
parser.error("Display ranks are not used without taxonomic info")
else:
if arguments.taxdir is None:
parser.error("Cannot select ranks without a taxonomy")
if arguments.ranks is None:
# set a default
arguments.ranks = [
'phylum',
'class',
'order',
'family',
'genus',
'species']
try:
# Make sure the rank lists make sense
arguments.ranks = cleanRanks(arguments.ranks)
if arguments.printRanks is not None:
arguments.printRanks = cleanRanks(arguments.printRanks)
except Exception as e:
parser.error(str(e))
# load weights file
sequenceWeights = loadSequenceWeights(arguments.weights)
# only print to stdout if there is a single rank
if len(arguments.ranks) > 1 and arguments.outfile is None:
parser.error("STDOUT only works if a single rank is chosen!")
# Because rank is used in parsing hits, we can only do multiple ranks for
# certain kinds of count methods
if len(arguments.ranks) > 1:
rank = None
if arguments.countMethod in ['consensus', 'most']:
parser.error(
"Using multiple ranks does not work with the 'consensus' "
"or 'most' counting methods. LCA should give the same "
"results as consensus. If you really want to do this, "
"use a bash loop:'for rank in phylum order genus; do "
"COMMAND -r ${rank}; done'")
else:
rank = arguments.ranks[0]
# load necessary maps
(taxonomy, hitStringMap) = readMaps(arguments)
# parse input files
fileCounts = {}
totals = {}
fileLabels = {}
sortedLabels = []
# Allow for file names to be preceded with TAG=
for filename in arguments.input_files:
bits = filename.split("=", 1)
if len(bits) > 1:
(filetag, filename) = bits
else:
filetag = filename
fileLabels[filename] = filetag
# keep order so that column order matches arguments
sortedLabels.append(filetag)
fileCounts[filetag] = {}
totals[filetag] = 0
params = FilterParams.create_from_arguments(arguments)
if arguments.countMethod == 'tophit' or arguments.countMethod == 'toporg':
# Process all files at once and use overall abundance to pick best hits
from edl import redistribute
multifile = redistribute.multipleFileWrapper(fileLabels.keys())
if arguments.countMethod == 'tophit':
# don't give any taxonomy, just map to accessions for
# redistribution
readHits = redistribute.pickBestHitByAbundance(
multifile,
filterParams=params,
returnLines=False,
winnerTakeAll=not arguments.proportional,
parseStyle=arguments.parseStyle,
sequenceWeights=sequenceWeights)
# define method to turn Hits into orgnaisms
hitTranslator = getHitTranslator(parseStyle=arguments.parseStyle,
taxonomy=taxonomy,
hitStringMap=hitStringMap)
translateHit = lambda hit: hitTranslator.translateHit(hit=hit)[0]
else:
# translate to organism before finding most abundant
readHits = redistribute.pickBestHitByAbundance(
multifile,
filterParams=params,
returnLines=False,
returnTranslations=True,
winnerTakeAll=not arguments.proportional,
taxonomy=taxonomy,
hitStringMap=hitStringMap,
parseStyle=ACCS)
# Organisms will be returned, make translator trivial:
translateHit = passThrough
# use read->file mapping and hit translator to get file based counts
# from returned (read,Hit) pairs
increment = 1
for (read_name, hit) in readHits:
file_name, read_name = read_name.split("/", 1)
file_tag = fileLabels[unquote_plus(file_name)]
taxon = translateHit(hit)
taxcount = fileCounts[file_tag].setdefault(taxon, 0)
if sequenceWeights is not None:
increment = sequenceWeights.get(read_name, 1)
fileCounts[file_tag][taxon] = taxcount + increment
totals[file_tag] += increment
logging.debug(str(totals))
else:
# Original way, just process each file separately
for (filename, filetag) in fileLabels.items():
infile = open(filename, 'rU')
hitIter = parseM8FileIter(infile,
hitStringMap,
params,
arguments.parseStyle,
arguments.countMethod,
taxonomy=taxonomy,
rank=rank)
(total, counts, hitMap) = \
countIterHits(hitIter,
allMethod=arguments.allMethod,
weights=sequenceWeights)
fileCounts[filetag] = counts
totals[filetag] = total
logging.info(
"parsed %d hits (%d unique) for %d reads from %s",
total, len(counts), len(hitMap), filename)
infile.close()
printCountTablesByRank(fileCounts, totals, sortedLabels, arguments)
def main():
description = __doc__
parser = argparse.ArgumentParser(description=description)
parser.add_argument("input_files", nargs="+",
default=[],
metavar="INFILE",
help="List of hit tables to process")
parser.add_argument("-o", "--outfile", dest="output_file",
metavar="OUTFILE", help="Write count table to OUTFILE")
parser.add_argument("-l", "--level", dest="levels", default=None,
metavar="LEVEL", action="append",
help=""" Level(s) to collect counts on. Use flag
multiple times to specify multiple levels. If multiple
values given, one table produced for each with rank
name appended to file name. Levels can be an integer
(1-3) for KEGG or SEED levels, any one of 'gene',
'role', 'family',
'ko', or 'ortholog' (which are all synonyms), or
anything not synonymous with 'gene' to
get CAZy groups. Defaults to ortholog/role and
levels 1, 2, and 3 for KEGG and SEED
and gene and group for CAZy and COG.""")
# option for deconvoluting clusters or assemblies
add_weight_arguments(parser, multiple=True)
# cutoff options
add_count_arguments(parser)
# format, ortholog heirarchy, and more
kegg.add_path_arguments(
parser,
defaults={'countMethod': 'tophit'},
choices={'countMethod':
('tophit',
'first',
'most',
'all',
'consensus')},
helps={'countMethod':
("How to deal with counts from multiple hits. ('first': "
"just use the first hit, 'most': "
"can return multiple hits, 'all': return every hit, "
"consensus: return None unless all the same). Do not "
"use most or consensus with more than one level at a time. "
"Default is 'tophit': This breaks any ties by choosing "
"the most abundant hit based on other unambiguous "
"assignments.")})
# log level and help
add_universal_arguments(parser)
arguments = parser.parse_args()
setup_logging(arguments)
if len(arguments.input_files) == 0:
parser.error("Must supply at least one m8 file to parse")
# Set defaults and check for some conflicts
if arguments.levels is None and arguments.heirarchyFile is None:
# using hit names only
arguments.levels = [None]
else:
if arguments.heirarchyFile is None \
and arguments.heirarchyType != 'cazy':
logging.warning("Type: %s", arguments.heirarchyType)
parser.error("Cannot select levels without a heirarchy (ko) file")
if arguments.levels is None:
# set a default
if arguments.heirarchyType is 'kegg':
arguments.levels = ['ko', '1', '2', 'pathway']
if arguments.heirarchyType is 'seed':
arguments.levels = ['role', '1', '2', 'subsystem']
else:
arguments.levels = ['gene', 'group']
try:
# Make sure the rank lists make sense
arguments.levels = cleanLevels(arguments.levels)
except Exception as e:
parser.error(str(e))
# load weights file
sequenceWeights = loadSequenceWeights(arguments.weights)
# only print to stdout if there is a single level
if len(arguments.levels) > 1 and arguments.output_file is None:
parser.error("STDOUT only works if a single level is chosen!")
cutoff = arguments.cutoff
# map reads to hits
if arguments.mapFile is not None:
if arguments.mapStyle == 'auto':
with open(arguments.mapFile) as f:
firstLine = next(f)
while len(firstLine) == 0 or firstLine[0] == '#':
firstLine = next(f)
if koMapRE.search(firstLine):
arguments.mapStyle = 'kegg'
elif seedMapRE.search(firstLine):
arguments.mapStyle = 'seed'
elif tabMapRE.search(firstLine):
arguments.mapStyle = 'tab'
# elif cogMapRE.search(firstLine):
# arguments.mapStyle='cog'
else:
raise Exception(
"Cannot figure out map type from first line:\n%s" %
(firstLine))
logging.info("Map file seems to be: %s", arguments.mapStyle)
if arguments.mapStyle == 'kegg':
valueMap = kegg.parseLinkFile(arguments.mapFile)
elif arguments.mapStyle == 'seed':
valueMap = kegg.parseSeedMap(arguments.mapFile)
# elif arguments.mapStyle=='cog':
# valueMap=kegg.parseCogMap(arguments.mapFile)
else:
if arguments.parseStyle == GIS:
keyType = int
else:
keyType = None
valueMap = parseMapFile(
arguments.mapFile,
valueType=None,
valueDelim=arguments.tab_map_delim,
keyType=keyType)
if len(valueMap) > 0:
logging.info("Read %d items into map. EG: %s",
len(valueMap), next(iter(valueMap.items())))
else:
logging.warn("Read 0 items into value map!")
else:
valueMap = None
# parse input files
fileCounts = {}
totals = {}
fileLabels = {}
sortedLabels = []
# Allow for file names to be preceded with TAG=
for filename in arguments.input_files:
bits = filename.split("=", 1)
if len(bits) > 1:
(filetag, filename) = bits
else:
filetag = filename
fileLabels[filename] = filetag
# keep order so that column order matches arguments
sortedLabels.append(filetag)
fileCounts[filetag] = {}
totals[filetag] = 0
params = FilterParams.create_from_arguments(arguments)
# TODO: incorporate weights into tophit algorithm!
if arguments.countMethod == 'tophit':
# Process all files at once and use overall abundance to pick best hits
from edl import redistribute
multifile = redistribute.multipleFileWrapper(fileLabels.items())
# don't give any hit translation, just use hit ids for redistribution
readHits = redistribute.pickBestHitByAbundance(
multifile,
filterParams=params,
returnLines=False,
winnerTakeAll=True,
parseStyle=arguments.parseStyle,
sequenceWeights=sequenceWeights)
# define method to turn Hits into Genes (kos, families)
hitTranslator = getHitTranslator(parseStyle=arguments.parseStyle,
hitStringMap=valueMap)
# translateHit = lambda hit: hitTranslator.translateHit(hit)[0]
# use read->file mapping and hit translator to get file based counts
# from returned (read,Hit) pairs
increment = 1
for (read_name, hit) in readHits:
file_tag, read_name = read_name.split("/", 1)
file_tag = unquote_plus(file_tag)
gene = hitTranslator.translateHit(hit)[0]
if gene is None:
gene = "None"
logging.debug(
"READ: %s\t%s\t%s\t%s",
file_tag, read_name, hit.hit, gene)
genecount = fileCounts[file_tag].setdefault(gene, 0)
if sequenceWeights is not None:
increment = sequenceWeights.get(read_name, 1)
fileCounts[file_tag][gene] = genecount + increment
totals[file_tag] += increment
logging.debug(str(totals))
else:
# Original way, just process each file separately
for (filename, filetag) in fileLabels.items():
infile = open(filename, 'rU')
hitIter = parseM8FileIter(infile,
valueMap,
params,
arguments.parseStyle,
arguments.countMethod,
ignoreEmptyHits=arguments.mappedHitsOnly)
(total, counts, hitMap) = \
countIterHits(hitIter,
allMethod=arguments.allMethod,
weights=sequenceWeights)
fileCounts[filetag] = counts
totals[filetag] = total
logging.info(
"parsed %d hits (%d unique) for %d reads from %s",
total, len(counts), len(hitMap), filename)
infile.close()
logging.debug(repr(fileCounts))
printCountTablesByLevel(fileCounts, totals, sortedLabels, arguments)
def main():
description = __doc__
parser = argparse.ArgumentParser(description)
add_IO_arguments(parser)
parser.add_argument("-l", "--level", dest="levels", default=None,
metavar="LEVEL", action="append",
help=""" Level(s) to collect counts on. Use flag
multiple times to specify multiple levels. If multiple
values given, one table produced for each with rank
name appended to file name. Levels can be an integer
(1-3) for KEGG or SEED levels, any one of 'gene',
'role', 'family',
'ko', or 'ortholog' (which are all synonyms), or
anything not synonymous with 'gene' to
get CAZy groups. Defaults to ortholog/role and
levels 1, 2, and 3 for KEGG and SEED
and gene and group for CAZy and COG.""")
parser.add_argument(
'-s',
'--squash',
dest='splitForLevels',
default=True,
action='store_false',
help="Don't split assignment rows if gene maps to multiple pathways, "
"just squash them into one row using python list syntax")
# format, ortholog heirarchy, and more
kegg.add_path_arguments(parser)
# log level and help
add_universal_arguments(parser)
arguments = parser.parse_args()
setup_logging(arguments)
# Set defaults and check for some conflicts
if arguments.levels is None and arguments.heirarchyFile is None:
# using hit names only
arguments.levels = [None]
else:
if arguments.heirarchyFile is None \
and arguments.heirarchyType != 'cazy':
logging.warn("Type: %s" % (arguments.heirarchyType))
parser.error("Cannot select levels without a heirarchy (ko) file")
if arguments.levels is None:
# set a default
if arguments.heirarchyType is 'kegg':
arguments.levels = ['ko', '1', '2', 'pathway']
if arguments.heirarchyType is 'seed':
arguments.levels = ['role', '1', '2', 'subsystem']
else:
arguments.levels = ['gene', 'group']
try:
# Make sure the level list makes sense
arguments.levels = cleanLevels(arguments.levels)
except Exception as e:
parser.error(str(e))
# map reads to hits
if arguments.mapFile is not None:
if arguments.mapStyle == 'auto':
with open(arguments.mapFile) as f:
firstLine = next(f)
while len(firstLine) == 0 or firstLine[0] == '#':
firstLine = next(f)
if koMapRE.search(firstLine):
arguments.mapStyle = 'kegg'
elif seedMapRE.search(firstLine):
arguments.mapStyle = 'seed'
elif tabMapRE.search(firstLine):
arguments.mapStyle = 'tab'
elif cogMapRE.search(firstLine):
arguments.mapStyle = 'cog'
else:
raise Exception(
"Cannot figure out map type from first line:\n%s" %
(firstLine))
logging.info("Map file seems to be: %s" % (arguments.mapStyle))
if arguments.mapStyle == 'kegg':
valueMap = kegg.parseLinkFile(arguments.mapFile)
elif arguments.mapStyle == 'seed':
valueMap = kegg.parseSeedMap(arguments.mapFile)
elif arguments.mapStyle == 'cog':
valueMap = kegg.parseCogMap(arguments.mapFile)
else:
if arguments.parseStyle == hits.GIS:
keyType = int
else:
keyType = None
valueMap = parseMapFile(
arguments.mapFile,
valueType=None,
keyType=keyType)
if len(valueMap) > 0:
logging.info("Read %d items into map. EG: %s" %
(len(valueMap), next(iter(valueMap.items()))))
else:
logging.warn("Read 0 items into value map!")
else:
valueMap = None
# set up level mapping
levelMappers = [getLevelMapper(l, arguments) for l in arguments.levels]
# parse input files
for (inhandle, outhandle) in inputIterator(arguments):
logging.debug(
"Reading from %s and writing to %s" %
(inhandle, outhandle))
hitMapIter = hits.parseM8FileIter(
inhandle,
valueMap,
arguments.hitTableFormat,
arguments.filterTopPct,
arguments.parseStyle,
arguments.countMethod,
ignoreEmptyHits=arguments.mappedHitsOnly)
if arguments.levels == [None]:
arguments.levels = ['Hit']
outhandle.write("Read\t%s\n" % ('\t'.join(arguments.levels)))
for read, hitIter in hitMapIter:
assignments = []
for hit in hitIter:
logging.debug("Hit: %s" % (hit))
assignment = []
for levelMapper in levelMappers:
assignment.append(levelMapper(hit))
assignments.append(assignment)
logging.debug("Read %s has %d hits" % (read, len(assignments)))
for assignment in assignments:
for assignmentList in handleMultipleMappings(
assignment, arguments):
outhandle.write(
"%s\t%s\n" %
(read, "\t".join(assignmentList)))
def main():
usage = "usage: %prog [OPTIONS] BLAST_M8_FILE[S]"
description = """
Takes a single m8 blast file and generates a table (or tables) of pathway/gene family assignments for the query sequences (aka 'reads'). Assignments can be for gene families, gene classes, or pathways. Multiple pathway or classification levels can be given. If they are, an assignment will be made at each level.
This differs from assignPathsToReadsFromBlast.py in that: (1) it can handle CAZy and SEED, (2) it will output multiple levels in one file, (3) multiple assignments are always printed on multiple lines.
This script will work with KEGG, SEED, or CAZy. CAZy only has one level of heirarchy, the others have 3. The CAZy heirarchy is apparent from the hit name and needs no supporting files. KEGG and SEED require mapping files to identify gene families and heirachy files to report levels other than the gene family or ortholog level. Both SEED and KEGG have three levels of classifications that can be indicated with a 1, 2, or 3. The words "subsystem" and "pathway" are synonyms for level 3.
If a count method is selected that can produce multiple assignments per read, each assignment will be printed on a new line.
NOTE: in KEGG (and SEED) a single ortholog (role) may belong to multiple pathways (subsystems). A hit to such an ortholog will result in extra assignment values for that query sequence (1 for each pathway it belongs to).
"""
parser = OptionParser(usage, description=description)
addIOOptions(parser)
parser.add_option("-l", "--level", dest="levels", default=None,
metavar="LEVEL", action="append",
help=""" Level(s) to collect counts on. Use flag
multiple times to specify multiple levels. If multiple
values given, one table produced for each with rank
name appended to file name. Levels can be an integer
(1-3) for KEGG or SEED levels, any one of 'gene', 'role', 'family',
'ko', or 'ortholog' (which are all synonyms), or
anything not synonymous with 'gene' to
get CAZy groups. Defaults to ortholog/role and
levels 1, 2, and 3 for KEGG and SEED
and gene and group for CAZy and COG.""")
parser.add_option('-s','--squash',dest='splitForLevels',
default=True, action='store_false',
help="Don't split assignment rows if gene maps to multiple pathways, just squash them into one row using python list syntax")
# format, ortholog heirarchy, and more
kegg.addPathOptions(parser)
# log level and help
addUniversalOptions(parser)
(options, args) = parser.parse_args()
setupLogging(options, description)
# Set defaults and check for some conflicts
if options.levels is None and options.heirarchyFile is None:
# using hit names only
options.levels=[None]
else:
if options.heirarchyFile is None and options.heirarchyType != 'cazy':
logging.warn("Type: %s" % (options.heirarchyType))
parser.error("Cannot select levels without a heirarchy (ko) file")
if options.levels is None:
# set a default
if options.heirarchyType is 'kegg':
options.levels=['ko','1','2','pathway']
if options.heirarchyType is 'seed':
options.levels=['role','1','2','subsystem']
else:
options.levels=['gene','group']
try:
# Make sure the level list makes sense
options.levels=cleanLevels(options.levels)
except Exception as e:
parser.error(str(e))
# only print to stdout if there is a single input file
if len(args)>1 and options.outfile is None:
parser.error("STDOUT only works if a single input file is given!")
# map reads to hits
if options.mapFile is not None:
if options.mapStyle == 'auto':
with open(options.mapFile) as f:
firstLine=f.next()
while len(firstLine)==0 or firstLine[0]=='#':
firstLine=f.next()
if koMapRE.search(firstLine):
options.mapStyle='kegg'
elif seedMapRE.search(firstLine):
options.mapStyle='seed'
elif tabMapRE.search(firstLine):
options.mapStyle='tab'
#elif cogMapRE.search(firstLine):
# options.mapStyle='cog'
else:
raise Exception("Cannot figure out map type from first line:\n%s" % (firstLine))
logging.info("Map file seems to be: %s" % (options.mapStyle))
if options.mapStyle=='kegg':
valueMap=kegg.parseLinkFile(options.mapFile)
elif options.mapStyle=='seed':
valueMap=kegg.parseSeedMap(options.mapFile)
#elif options.mapStyle=='cog':
# valueMap=kegg.parseCogMap(options.mapFile)
else:
if options.parseStyle == hits.GIS:
keyType=int
else:
keyType=None
valueMap = parseMapFile(options.mapFile,valueType=None,keyType=keyType)
if len(valueMap)>0:
logging.info("Read %d items into map. EG: %s" % (len(valueMap),valueMap.iteritems().next()))
else:
logging.warn("Read 0 items into value map!")
else:
valueMap=None
# set up level mapping
levelMappers = [getLevelMapper(l,options) for l in options.levels]
# parse input files
for (inhandle,outhandle) in inputIterator(args, options):
logging.debug("Reading from %s and writing to %s" % (inhandle, outhandle))
hitMapIter = hits.parseM8FileIter(inhandle, valueMap, options.hitTableFormat, options.filterTopPct, options.parseStyle, options.countMethod, ignoreEmptyHits=options.mappedHitsOnly,sortReads=options.hitTableSortReads)
outhandle.write("Read\t%s\n" % ('\t'.join(options.levels)))
for read, hitIter in hitMapIter:
assignments=[]
for hit in hitIter:
logging.debug("Hit: %s" % (hit))
assignment=[]
for levelMapper in levelMappers:
assignment.append(levelMapper(hit))
assignments.append(assignment)
logging.debug("Read %s has %d hits" % (read, len(assignments)))
for assignment in assignments:
for assignmentList in handleMultipleMappings(assignment,options):
outhandle.write("%s\t%s\n" % (read, "\t".join(assignmentList)))
def main():
description = __doc__
parser = argparse.ArgumentParser(description)
add_IO_arguments(parser)
parser.add_argument("-l",
"--level",
dest="levels",
default=None,
metavar="LEVEL",
action="append",
help=""" Level(s) to collect counts on. Use flag
multiple times to specify multiple levels. If multiple
values given, one table produced for each with rank
name appended to file name. Levels can be an integer
(1-3) for KEGG or SEED levels, any one of 'gene',
'role', 'family',
'ko', or 'ortholog' (which are all synonyms), or
anything not synonymous with 'gene' to
get CAZy groups. Defaults to ortholog/role and
levels 1, 2, and 3 for KEGG and SEED
and gene and group for CAZy and COG.""")
parser.add_argument(
'-S',
'--squash',
dest='splitForLevels',
default=True,
action='store_false',
help="Don't split assignment rows if gene maps to multiple pathways, "
"just squash them into one row using python list syntax")
# format, ortholog heirarchy, and more
kegg.add_path_arguments(parser)
# log level and help
add_universal_arguments(parser)
arguments = parser.parse_args()
setup_logging(arguments)
# Set defaults and check for some conflicts
if arguments.levels is None and arguments.heirarchyFile is None:
# using hit names only
arguments.levels = [None]
else:
if arguments.heirarchyFile is None \
and arguments.heirarchyType != 'cazy':
logging.warn("Type: %s" % (arguments.heirarchyType))
parser.error("Cannot select levels without a heirarchy (ko) file")
if arguments.levels is None:
# set a default
if arguments.heirarchyType is 'kegg':
arguments.levels = ['ko', '1', '2', 'pathway']
if arguments.heirarchyType is 'seed':
arguments.levels = ['role', '1', '2', 'subsystem']
else:
arguments.levels = ['gene', 'group']
try:
# Make sure the level list makes sense
arguments.levels = cleanLevels(arguments.levels)
except Exception as e:
parser.error(str(e))
# map reads to hits
if arguments.mapFile is not None:
if arguments.mapStyle == 'auto':
with open(arguments.mapFile) as f:
firstLine = next(f)
while len(firstLine) == 0 or firstLine[0] == '#':
firstLine = next(f)
if koMapRE.search(firstLine):
arguments.mapStyle = 'kegg'
elif seedMapRE.search(firstLine):
arguments.mapStyle = 'seed'
elif tabMapRE.search(firstLine):
arguments.mapStyle = 'tab'
elif cogMapRE.search(firstLine):
arguments.mapStyle = 'cog'
else:
raise Exception(
"Cannot figure out map type from first line:\n%s" %
(firstLine))
logging.info("Map file seems to be: %s" % (arguments.mapStyle))
if arguments.mapStyle == 'kegg':
valueMap = kegg.parseLinkFile(arguments.mapFile)
elif arguments.mapStyle == 'seed':
valueMap = kegg.parseSeedMap(arguments.mapFile)
elif arguments.mapStyle == 'cog':
valueMap = kegg.parseCogMap(arguments.mapFile)
else:
if arguments.parseStyle == hits.GIS:
keyType = int
else:
keyType = None
valueMap = parseMapFile(arguments.mapFile,
valueType=None,
valueDelim=arguments.tab_map_delim,
keyType=keyType)
if len(valueMap) > 0:
logging.info("Read %d items into map. EG: %s" %
(len(valueMap), next(iter(valueMap.items()))))
else:
logging.warn("Read 0 items into value map!")
else:
valueMap = None
# set up level mapping
levelMappers = [getLevelMapper(lvl, arguments) for lvl in arguments.levels]
# parse input files
for (inhandle, outhandle) in inputIterator(arguments):
logging.debug("Reading from %s and writing to %s" %
(inhandle, outhandle))
hitMapIter = hits.parseM8FileIter(
inhandle,
valueMap,
hits.FilterParams.create_from_arguments(arguments),
arguments.parseStyle,
arguments.countMethod,
ignoreEmptyHits=arguments.mappedHitsOnly)
if arguments.levels == [None]:
arguments.levels = ['Hit']
outhandle.write("Read\t%s\n" % ('\t'.join(arguments.levels)))
for read, hitIter in hitMapIter:
assignments = []
for hit in hitIter:
logging.debug("Hit: %s" % (hit))
assignment = []
for levelMapper in levelMappers:
assignment.append(levelMapper(hit))
assignments.append(assignment)
logging.debug("Read %s has %d hits" % (read, len(assignments)))
for assignment in assignments:
for assignmentList in handleMultipleMappings(
assignment, arguments):
outhandle.write("%s\t%s\n" %
(read, "\t".join(assignmentList)))
def main():
description = __doc__
parser = argparse.ArgumentParser(description=description)
parser.add_argument("input_files",
nargs="+",
default=[],
metavar="INFILE",
help="List of hit tables to process")
parser.add_argument("-o",
"--outfile",
dest="output_file",
metavar="OUTFILE",
help="Write count table to OUTFILE")
parser.add_argument("-l",
"--level",
dest="levels",
default=None,
metavar="LEVEL",
action="append",
help=""" Level(s) to collect counts on. Use flag
multiple times to specify multiple levels. If multiple
values given, one table produced for each with rank
name appended to file name. Levels can be an integer
(1-3) for KEGG or SEED levels, any one of 'gene',
'role', 'family',
'ko', or 'ortholog' (which are all synonyms), or
anything not synonymous with 'gene' to
get CAZy groups. Defaults to ortholog/role and
levels 1, 2, and 3 for KEGG and SEED
and gene and group for CAZy and COG.""")
# option for deconvoluting clusters or assemblies
add_weight_arguments(parser, multiple=True)
# cutoff options
add_count_arguments(parser)
# format, ortholog heirarchy, and more
kegg.add_path_arguments(
parser,
defaults={'countMethod': 'tophit'},
choices={
'countMethod': ('tophit', 'first', 'most', 'all', 'consensus')
},
helps={
'countMethod':
("How to deal with counts from multiple hits. ('first': "
"just use the first hit, 'most': "
"can return multiple hits, 'all': return every hit, "
"consensus: return None unless all the same). Do not "
"use most or consensus with more than one level at a time. "
"Default is 'tophit': This breaks any ties by choosing "
"the most abundant hit based on other unambiguous "
"assignments.")
})
# log level and help
add_universal_arguments(parser)
arguments = parser.parse_args()
setup_logging(arguments)
if len(arguments.input_files) == 0:
parser.error("Must supply at least one m8 file to parse")
# Set defaults and check for some conflicts
if arguments.levels is None and arguments.heirarchyFile is None:
# using hit names only
arguments.levels = [None]
else:
if arguments.heirarchyFile is None \
and arguments.heirarchyType != 'cazy':
logging.warning("Type: %s", arguments.heirarchyType)
parser.error("Cannot select levels without a heirarchy (ko) file")
if arguments.levels is None:
# set a default
if arguments.heirarchyType is 'kegg':
arguments.levels = ['ko', '1', '2', 'pathway']
if arguments.heirarchyType is 'seed':
arguments.levels = ['role', '1', '2', 'subsystem']
else:
arguments.levels = ['gene', 'group']
try:
# Make sure the rank lists make sense
arguments.levels = cleanLevels(arguments.levels)
except Exception as e:
parser.error(str(e))
# load weights file
sequenceWeights = loadSequenceWeights(arguments.weights)
# only print to stdout if there is a single level
if len(arguments.levels) > 1 and arguments.output_file is None:
parser.error("STDOUT only works if a single level is chosen!")
cutoff = arguments.cutoff
# map reads to hits
if arguments.mapFile is not None:
if arguments.mapStyle == 'auto':
with open(arguments.mapFile) as f:
firstLine = next(f)
while len(firstLine) == 0 or firstLine[0] == '#':
firstLine = next(f)
if koMapRE.search(firstLine):
arguments.mapStyle = 'kegg'
elif seedMapRE.search(firstLine):
arguments.mapStyle = 'seed'
elif tabMapRE.search(firstLine):
arguments.mapStyle = 'tab'
# elif cogMapRE.search(firstLine):
# arguments.mapStyle='cog'
else:
raise Exception(
"Cannot figure out map type from first line:\n%s" %
(firstLine))
logging.info("Map file seems to be: %s", arguments.mapStyle)
if arguments.mapStyle == 'kegg':
valueMap = kegg.parseLinkFile(arguments.mapFile)
elif arguments.mapStyle == 'seed':
valueMap = kegg.parseSeedMap(arguments.mapFile)
# elif arguments.mapStyle=='cog':
# valueMap=kegg.parseCogMap(arguments.mapFile)
else:
if arguments.parseStyle == GIS:
keyType = int
else:
keyType = None
valueMap = parseMapFile(arguments.mapFile,
valueType=None,
valueDelim=arguments.tab_map_delim,
keyType=keyType)
if len(valueMap) > 0:
logging.info("Read %d items into map. EG: %s", len(valueMap),
next(iter(valueMap.items())))
else:
logging.warn("Read 0 items into value map!")
else:
valueMap = None
# parse input files
fileCounts = {}
totals = {}
fileLabels = {}
sortedLabels = []
# Allow for file names to be preceded with TAG=
for filename in arguments.input_files:
bits = filename.split("=", 1)
if len(bits) > 1:
(filetag, filename) = bits
else:
filetag = filename
fileLabels[filename] = filetag
# keep order so that column order matches arguments
sortedLabels.append(filetag)
fileCounts[filetag] = {}
totals[filetag] = 0
# TODO: incorporate weights into tophit algorithm!
if arguments.countMethod == 'tophit':
# Process all files at once and use overall abundance to pick best hits
from edl import redistribute
params = FilterParams.create_from_arguments(arguments)
multifile = redistribute.multipleFileWrapper(fileLabels.items())
# don't give any hit translation, just use hit ids for redistribution
readHits = redistribute.pickBestHitByAbundance(
multifile,
filterParams=params,
returnLines=False,
winnerTakeAll=True,
parseStyle=arguments.parseStyle,
sequenceWeights=sequenceWeights)
# define method to turn Hits into Genes (kos, families)
hitTranslator = getHitTranslator(parseStyle=arguments.parseStyle,
hitStringMap=valueMap)
# translateHit = lambda hit: hitTranslator.translateHit(hit)[0]
# use read->file mapping and hit translator to get file based counts
# from returned (read,Hit) pairs
increment = 1
for (read_name, hit) in readHits:
file_tag, read_name = read_name.split("/", 1)
file_tag = unquote_plus(file_tag)
gene = hitTranslator.translateHit(hit)[0]
if gene is None:
gene = "None"
logging.debug("READ: %s\t%s\t%s\t%s", file_tag, read_name, hit.hit,
gene)
genecount = fileCounts[file_tag].setdefault(gene, 0)
if sequenceWeights is not None:
increment = sequenceWeights.get(read_name, 1)
fileCounts[file_tag][gene] = genecount + increment
totals[file_tag] += increment
logging.debug(str(totals))
else:
# Original way, just process each file separately
for (filename, filetag) in fileLabels.items():
infile = open(filename, 'rU')
hitIter = parseM8FileIter(infile,
valueMap,
arguments.hitTableFormat,
arguments.filter_top_pct,
arguments.parseStyle,
arguments.countMethod,
ignoreEmptyHits=arguments.mappedHitsOnly)
(total, counts, hitMap) = \
countIterHits(hitIter,
allMethod=arguments.allMethod,
weights=sequenceWeights)
fileCounts[filetag] = counts
totals[filetag] = total
logging.info("parsed %d hits (%d unique) for %d reads from %s",
total, len(counts), len(hitMap), filename)
infile.close()
logging.debug(repr(fileCounts))
printCountTablesByLevel(fileCounts, totals, sortedLabels, arguments)
def main():
"""" Set up the CLI """
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("input_files",
nargs="+",
default=[],
metavar="INFILE",
help="List of hit tables to process")
parser.add_argument("-o",
"--outfile",
dest="outfile",
metavar="OUTFILE",
help="Write count table to OUTFILE")
parser.add_argument("-r",
"--rank",
dest="ranks",
default=None,
metavar="RANK",
action="append",
help=""" Rank(s) to collect counts on. Use flag
multiple
times to specify multiple ranks. If multiple values
given, one table produced for each with rank name
appended to file name. Defaults to all major ranks
between phylum and species. Corresponds to rank names
in nodes.dmp. To see list run:
'cut -f5 nodes.dmp | uniq | sort | uniq'
in ncbi tax dir. Will also accept 'organism' to mean
no rank (ie, just the organism name).""")
parser.add_argument(
"-s",
"--collapseToDomain",
default=False,
action="store_true",
help="Collapse all taxa below given rank down to "
"superkingdom/domain. EG: in the genus output, anything "
"assigned to Cyanobactia, will be lumped in with all "
"other bacteria")
parser.add_argument(
"-R",
"--printRank",
dest="printRanks",
action="append",
help="Include indeicated rank(s) in lineage of printed taxa. "
"Will be ignored if beyond the rank of the taxa "
"(IE We can't include species if the taxon being counted "
"is genus)")
# option for deconvoluting clusters or assemblies
add_weight_arguments(parser, multiple=True)
# cutoff options
add_count_arguments(parser)
# format, tax dir, and more
add_taxon_arguments(parser,
choices={
'countMethod': ('LCA', 'all', 'first', 'most',
'tophit', 'toporg', 'consensus')
})
# log level and help
add_universal_arguments(parser)
arguments = parser.parse_args()
setup_logging(arguments)
if len(arguments.input_files) == 0:
parser.error("Must supply at least one m8 file to parse")
# Handle the case where Galaxy tries to set None as a string
arguments.ranks = checkNoneOption(arguments.ranks)
arguments.printRanks = checkNoneOption(arguments.printRanks)
logging.info("Printing out ranks: %r", arguments.ranks)
# Set defaults and check for some conflicts
if arguments.ranks is None and arguments.taxdir is None:
# using hit names only
arguments.ranks = [ORG_RANK]
if arguments.printRanks is not None:
parser.error("Display ranks are not used without taxonomic info")
else:
if arguments.taxdir is None:
parser.error("Cannot select ranks without a taxonomy")
if arguments.ranks is None:
# set a default
arguments.ranks = [
'phylum', 'class', 'order', 'family', 'genus', 'species'
]
try:
# Make sure the rank lists make sense
arguments.ranks = cleanRanks(arguments.ranks)
if arguments.printRanks is not None:
arguments.printRanks = cleanRanks(arguments.printRanks)
except Exception as e:
parser.error(str(e))
# load weights file
sequenceWeights = loadSequenceWeights(arguments.weights)
# only print to stdout if there is a single rank
if len(arguments.ranks) > 1 and arguments.outfile is None:
parser.error("STDOUT only works if a single rank is chosen!")
# Because rank is used in parsing hits, we can only do multiple ranks for
# certain kinds of count methods
if len(arguments.ranks) > 1:
rank = None
if arguments.countMethod in ['consensus', 'most']:
parser.error(
"Using multiple ranks does not work with the 'consensus' "
"or 'most' counting methods. LCA should give the same "
"results as consensus. If you really want to do this, "
"use a bash loop:'for rank in phylum order genus; do "
"COMMAND -r ${rank}; done'")
else:
rank = arguments.ranks[0]
# load necessary maps
(taxonomy, hitStringMap) = readMaps(arguments)
# parse input files
fileCounts = {}
totals = {}
fileLabels = {}
sortedLabels = []
# Allow for file names to be preceded with TAG=
for filename in arguments.input_files:
bits = filename.split("=", 1)
if len(bits) > 1:
(filetag, filename) = bits
else:
filetag = filename
fileLabels[filename] = filetag
# keep order so that column order matches arguments
sortedLabels.append(filetag)
fileCounts[filetag] = {}
totals[filetag] = 0
if arguments.countMethod == 'tophit' or arguments.countMethod == 'toporg':
# Process all files at once and use overall abundance to pick best hits
from edl import redistribute
params = FilterParams.create_from_arguments(arguments)
multifile = redistribute.multipleFileWrapper(fileLabels.keys())
if arguments.countMethod == 'tophit':
# don't give any taxonomy, just map to accessions for
# redistribution
readHits = redistribute.pickBestHitByAbundance(
multifile,
filterParams=params,
returnLines=False,
winnerTakeAll=True,
parseStyle=arguments.parseStyle,
sequenceWeights=sequenceWeights)
# define method to turn Hits into orgnaisms
hitTranslator = getHitTranslator(parseStyle=arguments.parseStyle,
taxonomy=taxonomy,
hitStringMap=hitStringMap)
translateHit = lambda hit: hitTranslator.translateHit(hit=hit)[0]
else:
# translate to organism before finding most abundant
readHits = redistribute.pickBestHitByAbundance(
multifile,
filterParams=params,
returnLines=False,
returnTranslations=True,
winnerTakeAll=True,
taxonomy=taxonomy,
hitStringMap=hitStringMap,
parseStyle=ACCS)
# Organisms will be returned, make translator trivial:
translateHit = passThrough
# use read->file mapping and hit translator to get file based counts
# from returned (read,Hit) pairs
increment = 1
for (read_name, hit) in readHits:
file_name, read_name = read_name.split("/", 1)
file_tag = fileLabels[unquote_plus(file_name)]
taxon = translateHit(hit)
taxcount = fileCounts[file_tag].setdefault(taxon, 0)
if sequenceWeights is not None:
increment = sequenceWeights.get(read_name, 1)
fileCounts[file_tag][taxon] = taxcount + increment
totals[file_tag] += increment
logging.debug(str(totals))
else:
# Original way, just process each file separately
for (filename, filetag) in fileLabels.items():
infile = open(filename, 'rU')
hitIter = parseM8FileIter(infile,
hitStringMap,
arguments.hitTableFormat,
arguments.filter_top_pct,
arguments.parseStyle,
arguments.countMethod,
taxonomy=taxonomy,
rank=rank)
(total, counts, hitMap) = \
countIterHits(hitIter,
allMethod=arguments.allMethod,
weights=sequenceWeights)
fileCounts[filetag] = counts
totals[filetag] = total
logging.info("parsed %d hits (%d unique) for %d reads from %s",
total, len(counts), len(hitMap), filename)
infile.close()
printCountTablesByRank(fileCounts, totals, sortedLabels, arguments)
def main():
""" The CLI """
description = """
Takes a hit table (reads searched against a database) and assigns
each read to a taxon. Hit table may be specified with -i or piped to STDIN.
Notes:
* Specifying a top score precent (-F) will force hits to be sorted
by score within each read. However, it is assumed that the hits in
the input table(s) are already grouped by read. This program does
not attempt to sort the entire input.
"""
parser = argparse.ArgumentParser(description)
util.add_IO_arguments(parser)
parser.add_argument("-T", "--taxids", default=False, action="store_true",
help="Output taxids instead of names")
edlhits.add_taxon_arguments(parser)
parser.add_argument(
"-r",
"--rank",
dest="rank",
default=None,
metavar="RANK",
help=" Rank to collect counts on. Defaults to None (whatever "
"the annotation was). Corresponds to rank names in nodes.dmp. "
"To see list run: 'cut -f5 nodes.dmp | uniq | sort | uniq' in "
"ncbi tax dir")
parser.add_argument(
"-R",
"--printRank",
dest="printRanks",
action="append",
help="Include indeicated rank(s) in lineage of printed taxa. "
"Will be ignored if beyond the rank of the taxa "
"(IE We can't include species if the taxon being counted "
"is genus)")
parser.add_argument(
"--no-header",
dest="no_header",
default=False,
action='store_true',
help="do not write header line")
util.add_universal_arguments(parser)
arguments = parser.parse_args()
util.setup_logging(arguments)
logging.debug("Parsing style is: %s", arguments.parseStyle)
# Handle the case where Galaxy tries to set None as a string
arguments.printRanks = util.checkNoneOption(arguments.printRanks)
# check arguments
if arguments.taxids and arguments.taxdir is None:
parser.error("Only use -T when a taxonomy is specified")
if arguments.rank is not None and arguments.taxdir is None:
parser.error(
"Please supply NCBI phylogeny(-n) if specifying a rank(-r).")
if arguments.printRanks is not None and arguments.taxdir is None:
parser.error(
"Please supply NCBI phylogeny(-n) if specifying a rank(-R).")
if arguments.rank is not None:
if arguments.rank == 'domain':
logging.warning('translating domain to superkingdom')
arguments.rank = 'superkingdom'
if arguments.rank not in ranks:
parser.error("Unknown rank: %s" % (arguments.rank))
try:
# Make sure the rank lists make sense
if arguments.printRanks is not None:
arguments.printRanks = cleanRanks(arguments.printRanks)
except Exception as exc:
parser.error(str(exc))
# load necessary maps
(taxonomy, value_map) = edlhits.readMaps(arguments)
# loop over inputs
for (inhandle, outhandle) in util.inputIterator(arguments):
logging.debug(
"Reading from %s and writing to %s",
inhandle, outhandle)
hit_iter = edlhits.parseM8FileIter(
inhandle,
value_map,
edlhits.FilterParams.create_from_arguments(arguments),
arguments.parseStyle,
arguments.countMethod,
taxonomy=taxonomy,
rank=arguments.rank)
##
# print output
# choose output method
if arguments.taxids:
hit_header = 'taxid'
printer = taxid_printer
else:
if arguments.printRanks is None:
hit_header = 'Hit(s)'
printer = default_printer
else:
hit_header = '\t'.join(arguments.printRanks)
def printer(read, hits):
" Inline function to reduce number of arguments "
return tax_table_printer(read,
hits,
arguments.rank,
arguments.printRanks)
# loop over reads
if not arguments.no_header:
outhandle.write("Read\t{}\n".format(hit_header))
for (read, hits) in hit_iter:
outhandle.write(printer(read, hits))
