def main(): parser = argparse.ArgumentParser(description=__doc__) parser.add_argument("contigs_fasta", help="The contigs as a fasta files") parser.add_argument("output_file_prefix", help="Three files are created with this prefix \ and one of hte suffixes: '.gff','.fna','.faa'") parser.add_argument("-r", "--rna-gff", action='append', default=[], help="GFF file containing RNA annotations. \ May be specified multiple times for \ multiple files") parser.add_argument("-c", "--cds-gff", action='append', default=[], help="GFF file containing RNA annotations. \ May be specified multiple times for \ multiple files") add_universal_arguments(parser) args = parser.parse_args() setup_logging(args) logging.info("Generating annotations for {contigs_fasta}". format(**vars(args))) merge_gffs(args.rna_gff, args.cds_gff, args.contigs_fasta, args.output_file_prefix)
def main(): description = __doc__ parser = argparse.ArgumentParser(description) parser.add_argument("-t", "--taxfile", dest="taxfile", metavar="FILE", help="Read Silva ranks from FILE") parser.add_argument("-d","--dbType",default="silva",choices=['silva','pr2'], help="Which database are we importing: 'silva' (default) or 'pr2'") parser.add_argument("-f","--fastaout",default=None, metavar='FILE', help="Write fasta with modified headers to FILE. Only used for pr2") parser.add_argument("-i","--idStart", default=0, type=int, help="Start taxid counting with this number") parser.add_argument("-o", "--output_map", default="lastdb.tax", help="File (relative to OUTPUT_DIR) to write id->taxid map to. Defaults to lastdb.tax") parser.add_argument('silva_fasta', nargs=1, metavar='SILVA_FASTA') parser.add_argument('output_dir', nargs=1, metavar="OUTPUT_DIR", help="Directory in which to create files") # logging and help add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) # name logger logger=logging.getLogger(sys.argv[0]) fastafile = arguments.silva_fasta[0] dumpDir = arguments.output_dir[0] # parse the input files if arguments.dbType=='pr2': rootNode, taxmap = buildPR2Tree(fastafile, fastaout=arguments.fastaout, nextId=arguments.idStart, ) else: if arguments.taxfile is None: parser.error("You must provide the tax file for Silva (-t)") rootNode, taxmap = buildSilvaTree(arguments.taxfile, fastafile, logger) nodesFile = os.path.sep.join((dumpDir, 'nodes.dmp')) namesFile = os.path.sep.join((dumpDir, 'names.dmp')) taxFile = os.path.sep.join((dumpDir, arguments.output_map)) logger.info("Writing taxonomy to %s and %s" % (nodesFile, namesFile)) with open(nodesFile,'w') as nodeFile: with open(namesFile,'w') as nameFile: writeDumpFiles(rootNode, nodeFile, nameFile) logger.info("Writing taxid table to %s" % (taxFile)) with open(taxFile,'w') as taxMapFile: for (hitid, taxid) in taxmap.items(): taxMapFile.write("%s\t%s\n" % (hitid, taxid))
def main(): """ Sets up the command line interface """ description = __doc__ parser = argparse.ArgumentParser(description=description) parser.add_argument('infile', nargs='?', type=argparse.FileType('rU'), default=sys.stdin, help=("Input file (defaults to STDIN) containing " "a value on each line")) parser.add_argument('outfile', nargs='?', type=argparse.FileType('w'), default=sys.stdout, help="File to which to write histogram") parser.add_argument('-b', '--bins', type=int, default=50) parser.add_argument('-l', '--label', default="value") parser.add_argument('-w', '--width', default=75, type=int) parser.add_argument('-L', '--log', action='store_true') parser.add_argument('-W', '--max-label-width', type=int, default=10) # log level and help add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) values = [] for line in arguments.infile: try: values.append(float(line.strip())) except ValueError: if len(line.strip()) > 0: logging.warning("Skipping bad line:\n%s", line.strip()) logging.info("Read in %d values", len(values)) arguments.outfile.write( ascii_histogram( histogram(values, bins=arguments.bins), label=arguments.label, width=arguments.width, log=arguments.log, maxLabelWidth=arguments.max_label_width, ))
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))
def main(): # set up CLI description = """ Takes a tabular text file and translate a column of KO values into a new column of KEGG pathways. KO column can have multiple entries per row. Output column will have multiple entries per pathway cell. """ parser = argparse.ArgumentParser(description=description) util.add_IO_arguments(parser) parser.add_argument("-l", "--level", dest="level", default="PATHWAY", metavar="LEVEL", help=""" Level to collect counts on. Level can be one of: NAME, PATHWAY, EC, DEFINITION, or a level in the CLASS heirachy: 1, 2, or 3 """) parser.add_argument( "-f", "--fill_missing", dest="fill", metavar="FILL", default="No Pathway", help="Put FILL in column when KO has no pathway assigned. " "Defaults to 'No Pathway'") # format, ortholog heirarchy, and more parser.add_argument("-k", "--ko_file", dest="ko_file", metavar="MAPFILE", help="Location of kegg ko file") parser.add_argument("-c", "--ko_column", type=int, default=1, help="Column number (first column is 1)", metavar="COLUMN") parser.add_argument( "-C", "--new_column", type=int, default=None, help="Column number to insert new column after. Default is the " "after the source column. 0=>make it the first column. " "-1=>make it the last column.", metavar="COLUMN") parser.add_argument( "-L", "--long_output", default=False, action='store_true', help="Insert new duplicate rows if KO maps to multiple values") parser.add_argument( "-H", "--header", default=None, metavar='HEADER', help="Put HEADER in first row instead of trying to translate") parser.add_argument("-Q", "--quotes", default=False, action="store_true", help="Encase translated values in double quotes") parser.add_argument( "-s", "--sep", dest='sep', default='\t', help="""Character separating table cells. Default is tab""") parser.add_argument( "-S", "--ko_sep", dest='ko_sep', default=';', help="""Character separating multiple KO values in iput table and used to separate multiple values in output column. Default is ";". Ignored for output if --longOutput requested""") # log level and help util.add_universal_arguments(parser) arguments = parser.parse_args() util.setup_logging(arguments) logging.info("KO mapping from: " + arguments.ko_file) logging.debug("Fill: '%s'" % (arguments.fill)) translation = kegg.parse_KEGG_file(arguments.ko_file, arguments.level) # switch to zero indexing if arguments.new_column: arguments.new_column -= 1 arguments.ko_column -= 1 for (inhandle, outhandle) in util.inputIterator(arguments): for new_line in translate_ko_column( inhandle, sep=arguments.sep, ko_sep=arguments.ko_sep, ko_column=arguments.ko_column, new_column=arguments.new_column, translation=translation, default=arguments.fill, quotes=arguments.quotes, header=arguments.header, long_out=arguments.long_output, ): outhandle.write(new_line)
def main(): description = __doc__ # command line options parser = argparse.ArgumentParser(description, conflict_handler='resolve') parser.add_argument("input_files", nargs=1, default=[], metavar="INFILE", help="Hit table to process") parser.add_argument( "-o", "--outfile", dest="outfile", metavar="OUTFILE", help="Write masked fasta output to OUTFILE (default is STDOUT).") parser.add_argument( "-i", "--infile", dest="fasta", metavar="FILE", help=" File containing the fasta (defaults to STDIN)") parser.add_argument( "-M", "--mask", dest="keep", default=True, action="store_false", help="Return unmatched sequence fragments instead of hits.") parser.add_argument("-m", "--minLength", dest="minLength", type=int, metavar="BASES", default=1, help="minimum number of bases for sequences in output") parser.add_argument( "-n", "--numbering_prefix", default=None, help="If given, name extracted sequence with this scring followed " "by a sinmple counting index of all extracted sequences. For " "example, -n \"r\" would add _r1 to the end of the first " "extracted sequence and _r2 to the second, and so on. By " "default, extracted sequences are named with start_end " "positions.") parser.add_argument( "-t", "--translate", default=False, action='store_true', help="Transalte to Amino Acid sequences") add_hit_table_arguments(parser, flags='all') # log level and help add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) # check that we have blast file as argument if len(arguments.input_files) != 1: parser.error( "Please supply the name of a hit table as the only argument") blastFile = arguments.input_files[0] # set up input/output streams if arguments.fasta is None: fastaHandle = sys.stdin fastaStr = 'STDIN' else: fastaHandle = open(arguments.fasta, "rt") fastaStr = arguments.fasta logging.info( "Extrating sequence fragments from %s based on hits in %s" % (fastaStr, blastFile)) if arguments.outfile is None: logging.info("Writing %s sequences to STDOUT" % ('fasta')) outputHandle = sys.stdout else: logging.info( "Writing %s sequences to %s" % ('fasta', arguments.outfile)) outputHandle = open(arguments.outfile, 'w') # load hit regions if arguments.keep: minHitLength = arguments.minLength else: minHitLength = 1 readHits = loadHitRegions(blastFile, minHitLength, arguments) logging.info("Found hits for %d reads" % (len(readHits))) # process the fasta file with hits extractHits( fastaHandle, outputHandle, readHits, arguments.translate, arguments.minLength, arguments.keep, arguments.numbering_prefix)
def main(): """ set up the command line interface """ parser = argparse.ArgumentParser(description=__doc__) parser.add_argument("-1", "--input_file_1", default=None, type=argparse.FileType('r'), metavar=("INPUT_TABLE_1"), help="Input table 1") parser.add_argument("-2", "--input_file_2", default=None, type=argparse.FileType('r'), metavar=("INPUT_TABLE_2"), help="Input table 2") parser.add_argument("-m", "--multiplier", default=None, metavar=("MULTIPLIER_TABLE"), help=("Table of values to multiply each sequence. " "EG assembly coverages.")) parser.add_argument("-T", "--total_reads", default=0, metavar="TOTAL_READS", type=int, help="Total number of reads to expect. (This allows " "the reporting of unknown read count)") parser.add_argument( "-o", "--outfile", dest="outfile", type=argparse.FileType('w'), default=sys.stdout, metavar="OUTFILE", help="Write count table to OUTFILE. (Defaults to STDOUT") parser.add_argument( "-L", "--long_output", default=False, action="store_true", help="Print one number per row (prefixed by two keys) instead " "of a table with one seet of keys as column names and one " "set as row names.") parser.add_argument( "-H", "--hitCol1", dest="hitCol1", type=int, default=-1, help="Index (starting at 0) of column in file 1 with hit name, -1 " "is default meaning all columns that are not the read name are " "hit names.", metavar="HITCOL") parser.add_argument( "-I", "--hitCol2", dest="hitCol2", type=int, default=-1, help="Index (starting at 0) of column in file 2 with hit name, -1 " "is default meaning all columns that are not the read name " "are hit names.", metavar="HITCOL") parser.add_argument( "-S", "--skipFirstRow", action="store_true", default=False, help="hit tables have a header row which needs to be skipped") add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) if arguments.input_file_1 is None or arguments.input_file_2 is None: parser.error("Please supply two input files") logging.info("reading hits from %s", arguments.input_file_1.name) hits1 = parseHits(arguments.input_file_1, 0, arguments.hitCol1, arguments.skipFirstRow, None) logging.info("reading hits from %s", arguments.input_file_2.name) hits2 = parseHits(arguments.input_file_2, 0, arguments.hitCol2, arguments.skipFirstRow, None) hits1 = tupleIteratorToMap(hits1) hits2 = tupleIteratorToMap(hits2) if arguments.multiplier is not None: multipliers = parseMapFile(arguments.multiplier, valueType=float) else: multipliers = None logging.info("counting hits") (table, cols) = combine_counts(hits1, hits2, multipliers, total_reads=arguments.total_reads) # print out hit table logging.info("printing table to " + arguments.outfile.name) print_table(arguments.outfile, table, cols, is_multiplied=multipliers is not None, long_output=arguments.long_output)
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(): 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(): # set up CLI description = __doc__ parser = argparse.ArgumentParser(description=description) parser.add_argument("-i", "--infile", dest="infile", metavar="FILE", help="Read raw table from INFILE") parser.add_argument( "-o", "--outfile", dest="outfile", metavar="OUTFILE", help="Write collapsed table to OUTFILE") parser.add_argument("-d", "--delim", dest="delim", default="\t", help="Input table delimiter", metavar="DELIM") parser.add_argument("-D", "--delimOut", dest="delimOut", default="\t", help="Output table delimiter", metavar="DELIM") parser.add_argument( '-F', '--countFirst', action='store_true', default=False, help="Don't skip the first line, it's NOT a header") parser.add_argument( "-R", "--readColumn", dest="readCol", type=int, default=0, help="Index (starting at 0) of column with read name, 0 is default", metavar="READCOL") parser.add_argument( "-H", "--hitColumn", dest="hitCol", type=int, default=2, help="Index (starting at 0) of column with hit name (for counting), " "2 is default, if less than zero, all (non-read) columns will " "be used as multiple hits", metavar="HITCOL") parser.add_argument( '-s', '--hitSep', default=None, help="Use this string to split multiple values in single hit cell. " "Default is 'None' to leave hits as is, use 'eval' to parse " "as python repr strings") add_weight_arguments(parser, multiple=False) parser.add_argument("-T", "--total", default=False, action="store_true", help="Report 'Total' in the first row") # cutoff options add_count_arguments(parser, {'cutoff': 0}) # log level and help add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) # make sure we have something to do if (arguments.infile is None): logging.info("Reading table from: STDIN") else: logging.info("Reading table from: " + arguments.infile) if (arguments.outfile is None): logging.info("Writing counts to: STDOUT") else: logging.info("Writing counts to: " + arguments.outfile) # process arguments takeFirst = (arguments.allMethod == 'first') splitHits = (arguments.hitSep is not None and arguments.hitSep != 'None') uncluster = (arguments.weights is not None) if arguments.hitSep == 'eval': parser.error("Sorry, parsing with eval is not yet supported!") # inform the curious user logging.info("Delimiter: '" + arguments.delim) logging.info("Read names in col: '" + str(arguments.readCol)) logging.info("Hit names in col: '" + str(arguments.hitCol)) if splitHits: logging.info("Splitting hits with: %s" % (arguments.hitSep)) logging.warn( "Splitting hits has not been tested yet! Let me know how it goes.") if takeFirst: logging.info("Taking first hit for each read.") else: if arguments.allMethod == 'portion': logging.info("Dividing count among all hits for each read.") else: logging.info("Adding 1 to every hit for each read") if uncluster: logging.info( "Getting read cluster sizes from: %s" % (arguments.weights)) if arguments.countFirst: logging.info("First line is data") else: logging.info("Skipping first line") # Do the counting! counts = {} countHitsForRead = getAllMethod(arguments.allMethod) clusteredReadCounts = {} if uncluster: clusteredReadCounts = parseMapFile( arguments.clusterFile, valueType=int) currentRead = '' readCount = 1 hits = [] if arguments.infile is None: infile = sys.stdin else: infile = open(arguments.infile) # loop over lines if not arguments.countFirst: # skip first line try: next(infile) except StopIteration: raise Exception("No lines in %s" % str(infile)) for line in infile: line = line.rstrip('\r\n') rowcells = line.split(arguments.delim) # get read read = rowcells[arguments.readCol] # if it's a new read, process previous read if currentRead == '': currentRead = read elif read != currentRead and currentRead != '': readCount += 1 logging.info("Checking hits for %s" % currentRead) # was it part of a cluster? multiplier = 1 if uncluster: multiplier = clusteredReadCounts[currentRead] # where does the count for this read go countHitsForRead(hits, counts, multiplier=multiplier) hits = [] currentRead = read # get hit from this line if arguments.hitCol >= 0: hit = rowcells[arguments.hitCol] if splitHits: hits.extend(hit.split(arguments.hitSep)) else: hits.append(hit) else: rowcells.pop(arguments.readCol) hits.extend(rowcells) # check last read! logging.info("Checking hits for %s" % currentRead) # was it part of a cluster? multiplier = 1 if uncluster: multiplier = clusteredReadCounts[currentRead] # where does the count for this read go countHitsForRead(hits, counts, multiplier=multiplier) # apply cutoff if arguments.cutoff > 0: applyFractionalCutoff(counts, threshold=arguments.cutoff * readCount) # print output if arguments.outfile is None: outhandle = sys.stdout else: outhandle = open(arguments.outfile, 'w') if arguments.total: outhandle.write("Total%s%d\n" % (arguments.delimOut, readCount)) if arguments.allMethod == 'portion': outFmtString = "%s%s%f\n" else: outFmtString = "%s%s%d\n" delimRE = re.compile(arguments.delimOut) for hit in sorted(counts.keys()): count = counts[hit] hit = delimRE.sub('_', hit) outhandle.write(outFmtString % (hit, arguments.delimOut, count))
def main(): description = """ Take a blast result table and output a subset of hits based on the chosen filtering options. If more than one blast file given, use -O to get multiple output files, otherwise all output data will be concatenated into one output. """ # command line arguments parser = argparse.ArgumentParser(description=description, conflict_handler='resolve') add_hit_table_arguments(parser, flags='all') parser.add_argument("-o", "--outfilenome", dest="outfilename", default=None, metavar="OUTFILENAME", help="Write masked fasta output to OUTFILENAME.") parser.add_argument( '-O', '--autoOutName', default=False, action='store_true', help="Automatically generate output file name from input name " "and options. Overridden by -o, cannot be used with data " "from STDIN.") parser.add_argument('-G', '--gff', default=False, action='store_true', help="output GFF format instead of input format") parser.add_argument('hit_table', nargs='*', type=argparse.FileType('rU'), default=[ sys.stdin, ], help="Table of search results to be filtered. " "If absent, data will be read from STDIN") add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) # check that we have blast file as argument # if we're not doing auto file names, wriate all outputs to same file if not arguments.autoOutName: if arguments.outfilename is not None: logging.info("Writing data to %s" % (arguments.outfilename)) outfile_handle = open(arguments.outfilename, 'w') else: logging.info("writing data to STDOUT") outfile_handle = sys.stdout if arguments.gff: logging.info("Converting to GFF") # loop over inputs for infile_handle in arguments.hit_table: logging.info("reading data from %s" % (infile_handle.name)) if arguments.autoOutName: outfile_handle = open(getOutputFile(infile_handle.name, arguments), 'w') # filter params = FilterParams.create_from_arguments(arguments) filterM8(infile_handle, outfile_handle, params, to_gff=arguments.gff) if arguments.autoOutName: outfile_handle.close() infile_handle.close()
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(): 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(): # command line arguments parser = argparse.ArgumentParser( description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter, conflict_handler='resolve') # default to non-overlapping=0 add_hit_table_arguments(parser, flags='all', defaults={'nonoverlapping': 0}) parser.add_argument("-o", "--outfilenome", dest="outfilename", default=None, metavar="OUTFILENAME", help="Write masked fasta output to OUTFILENAME.") parser.add_argument('hit_table', nargs='?', type=argparse.FileType('rU'), default=sys.stdin, help="Table of search results to be filtered. " "If absent, data will be read from STDIN") add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) # output file or STDOUT if arguments.outfilename is not None: logging.info("Writing data to %s" % (arguments.outfilename)) outfile_handle = open(arguments.outfilename, 'w') else: logging.info("writing data to STDOUT") outfile_handle = sys.stdout # input file or STDIN (handled by argparse) infile_handle = arguments.hit_table logging.info("reading data from %s" % (infile_handle.name)) # filter, but don't apply nonoverlapping yet # non-overlapping should be applied per-reference only params = FilterParams.create_from_arguments(arguments) # save user supplied value for later overlap_buffer = params.nonoverlapping # turn off for now params.set_nonoverlapping(-1) # merge hit_iter = filterM8Stream(infile_handle, params, return_lines=False) for query, query_hits in hit_iter: # group by reference hit hits_by_ref = defaultdict(list) for hit in query_hits: hits_by_ref[hit.hit].append(hit) # one output for query/reference pair for ref, ref_hits in hits_by_ref.items(): # remove overlaps unless the buffer has been set to <0 if overlap_buffer >= 0: ref_hits = remove_overlapping_hits( ref_hits, on_hit=True, buffer=params.nonoverlapping) ref_hits = remove_overlapping_hits( ref_hits, on_hit=False, buffer=params.nonoverlapping) # aggregate values length, score, identities = 0, 0, 0 for hit in ref_hits: length += hit.mlen score += hit.score try: # this will be off by 100x identities += hit.pctid * hit.mlen except: # just report pctid=0 if no pctid column in input pass outfile_handle.write( "%s\t%s\t%d\t%d\t%0.2f\n" % (query, ref, length, score, identities / length)) outfile_handle.close() infile_handle.close()
def main(): description = __doc__ # command line options parser = argparse.ArgumentParser(description, conflict_handler='resolve') parser.add_argument("input_files", nargs=1, default=[], metavar="INFILE", help="Hit table to process") parser.add_argument( "-o", "--outfile", dest="outfile", metavar="OUTFILE", help="Write masked fasta output to OUTFILE (default is STDOUT).") parser.add_argument( "-i", "--infile", dest="fasta", metavar="FILE", help=" File containing the fasta (defaults to STDIN)") parser.add_argument( "-M", "--mask", dest="keep", default=True, action="store_false", help="Return unmatched sequence fragments instead of hits.") parser.add_argument("-m", "--minLength", dest="minLength", type=int, metavar="BASES", default=1, help="minimum number of bases for sequences in output") parser.add_argument( "-n", "--numbering_prefix", default=None, help="If given, name extracted sequence with this scring followed " "by a sinmple counting index of all extracted sequences. For " "example, -n \"r\" would add _r1 to the end of the first " "extracted sequence and _r2 to the second, and so on. By " "default, extracted sequences are named with start_end " "positions.") parser.add_argument( "-t", "--translate", default=False, action='store_true', help="Transalte to Amino Acid sequences") add_hit_table_arguments(parser, flags='all') # log level and help add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) # check that we have blast file as argument if len(arguments.input_files) != 1: parser.error( "Please supply the name of a hit table as the only argument") blastFile = arguments.input_files[0] # set up input/output streams if arguments.fasta is None: fastaHandle = sys.stdin fastaStr = 'STDIN' else: fastaHandle = open(arguments.fasta, "rU") fastaStr = arguments.fasta logging.info( "Extrating sequence fragments from %s based on hits in %s" % (fastaStr, blastFile)) if arguments.outfile is None: logging.info("Writing %s sequences to STDOUT" % ('fasta')) outputHandle = sys.stdout else: logging.info( "Writing %s sequences to %s" % ('fasta', arguments.outfile)) outputHandle = open(arguments.outfile, 'w') # load hit regions if arguments.keep: minHitLength = arguments.minLength else: minHitLength = 1 readHits = loadHitRegions(blastFile, minHitLength, arguments) logging.info("Found hits for %d reads" % (len(readHits))) # process the fasta file with hits extractHits( fastaHandle, outputHandle, readHits, arguments.translate, arguments.minLength, arguments.keep, arguments.numbering_prefix)
def main(): """ set up the command line interface """ parser = argparse.ArgumentParser(description=__doc__) parser.add_argument("-1", "--input_file_1", default=None, type=argparse.FileType('r'), metavar=("INPUT_TABLE_1"), help="Input table 1") parser.add_argument("-2", "--input_file_2", default=None, type=argparse.FileType('r'), metavar=("INPUT_TABLE_2"), help="Input table 2") parser.add_argument("-m", "--multiplier", default=None, metavar=("MULTIPLIER_TABLE"), help=("Table of values to multiply each sequence. " "EG assembly coverages.")) parser.add_argument("-T", "--total_reads", default=0, metavar="TOTAL_READS", type=int, help="Total number of reads to expect. (This allows " "the reporting of unknown read count)") parser.add_argument( "-o", "--outfile", dest="outfile", type=argparse.FileType('w'), default=sys.stdout, metavar="OUTFILE", help="Write count table to OUTFILE. (Defaults to STDOUT") parser.add_argument( "-L", "--long_output", default=False, action="store_true", help="Print one number per row (prefixed by two keys) instead " "of a table with one seet of keys as column names and one " "set as row names.") parser.add_argument( "-H", "--hitCol1", dest="hitCol1", type=int, default=-1, help="Index (starting at 0) of column in file 1 with hit name, -1 " "is default meaning all columns that are not the read name are " "hit names.", metavar="HITCOL") parser.add_argument( "-I", "--hitCol2", dest="hitCol2", type=int, default=- 1, help="Index (starting at 0) of column in file 2 with hit name, -1 " "is default meaning all columns that are not the read name " "are hit names.", metavar="HITCOL") parser.add_argument( "-S", "--skipFirstRow", action="store_true", default=False, help="hit tables have a header row which needs to be skipped") add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) if arguments.input_file_1 is None or arguments.input_file_2 is None: parser.error("Please supply two input files") logging.info("reading hits from %s", arguments.input_file_1.name) hits1 = parseHits(arguments.input_file_1, 0, arguments.hitCol1, arguments.skipFirstRow, None) logging.info("reading hits from %s", arguments.input_file_2.name) hits2 = parseHits(arguments.input_file_2, 0, arguments.hitCol2, arguments.skipFirstRow, None) hits1 = tupleIteratorToMap(hits1) hits2 = tupleIteratorToMap(hits2) if arguments.multiplier is not None: multipliers = parseMapFile(arguments.multiplier, valueType=float) else: multipliers = None logging.info("counting hits") (table, cols) = combine_counts(hits1, hits2, multipliers, total_reads=arguments.total_reads) # print out hit table logging.info("printing table to " + arguments.outfile.name) print_table(arguments.outfile, table, cols, is_multiplied=multipliers is not None, long_output=arguments.long_output)
def main(): description = """ Take a blast result table and output a subset of hits based on the chosen filtering options. If more than one blast file given, use -O to get multiple output files, otherwise all output data will be concatenated into one output. """ # command line arguments parser = argparse.ArgumentParser( description=description, conflict_handler='resolve') add_hit_table_arguments(parser, flags='all') parser.add_argument( "-o", "--outfilenome", dest="outfilename", default=None, metavar="OUTFILENAME", help="Write masked fasta output to OUTFILENAME.") parser.add_argument( '-O', '--autoOutName', default=False, action='store_true', help="Automatically generate output file name from input name " "and options. Overridden by -o, cannot be used with data " "from STDIN.") parser.add_argument('-G', '--gff', default=False, action='store_true', help="output GFF format instead of input format") parser.add_argument('hit_table', nargs='*', type=argparse.FileType('rU'), default=[sys.stdin, ], help="Table of search results to be filtered. " "If absent, data will be read from STDIN") add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) # check that we have blast file as argument # if we're not doing auto file names, wriate all outputs to same file if not arguments.autoOutName: if arguments.outfilename is not None: logging.info("Writing data to %s" % (arguments.outfilename)) outfile_handle = open(arguments.outfilename, 'w') else: logging.info("writing data to STDOUT") outfile_handle = sys.stdout if arguments.gff: logging.info("Converting to GFF") # loop over inputs for infile_handle in arguments.hit_table: logging.info("reading data from %s" % (infile_handle.name)) if arguments.autoOutName: outfile_handle = open( getOutputFile( infile_handle.name, arguments), 'w') # filter params = FilterParams.create_from_arguments(arguments) filterM8(infile_handle, outfile_handle, params, to_gff=arguments.gff) if arguments.autoOutName: outfile_handle.close() infile_handle.close()
def main(): description = __doc__ parser = argparse.ArgumentParser(description=description) add_IO_arguments(parser) add_taxon_arguments( parser, defaults={ 'filter_top_pct': 0, 'parseStyle': ACCS, 'countMethod': 'tophit'}, choices={ 'countMethod': ( 'tophit', 'toporg')}) parser.add_argument( "-P", "--proportional", dest="proportional", default=False, action="store_true", help="Assign reads that have multiple equal top hits to taxa such " "that the overal proportion of taxa is consistent with the " "unambiguious hits. This is meant for use with the 'toporg' " "count method.") parser.add_argument( "-i", "--individualFiles", dest="individual", default=False, action="store_true", help="Use this flag to process files independently. Normally, " "counts from all files are pooled for making choices.") add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) # load necessary maps params = FilterParams.create_from_arguments(arguments) if arguments.countMethod == 'toporg': (taxonomy, hitStringMap) = readMaps(arguments) wta = not(arguments.proportional) if len(arguments.input_files) <= 1 or arguments.individual: # loop over input for (inhandle, outhandle) in inputIterator(arguments): logging.debug( "Reading from %s and writing to %s" % (inhandle, outhandle)) m8stream = M8Stream(inhandle) if arguments.countMethod == 'tophit': # don't give any taxonomy, just map to accessions for # redistribution readHits = redistribute.pickBestHitByAbundance( m8stream, filterParams=params, returnLines=True, winnerTakeAll=wta, parseStyle=arguments.parseStyle) else: # translate to organism before finding most abundant readHits = redistribute.pickBestHitByAbundance( m8stream, filterParams=params, returnLines=True, winnerTakeAll=wta, taxonomy=taxonomy, hitStringMap=hitStringMap, parseStyle=arguments.parseStyle) for line in readHits: outhandle.write(line) else: # process all files at once multifile = redistribute.multipleFileWrapper(arguments.input_files) # Build a map from input file name to output handle outputMap = {} for infile_handle in arguments.input_files: infile_name = infile_handle.name if arguments.output_file is None: outputMap[infile_name] = sys.stdout elif len(arguments.input_files) <= 1: outputMap[infile_name] = open(arguments.output_file, 'w') else: # use outfileName as suffix if arguments.cwd: # strip path info first (infilePath, infileFile) = os.path.split(infile_name) outfile = "./" + infileFile + arguments.output_file else: outfile = infile_name + arguments.output_file outputMap[infile_name] = open(outfile, 'w') if arguments.countMethod == 'tophit': # don't give any taxonomy, just map to accessions for # redistribution readHits = redistribute.pickBestHitByAbundance( multifile, filterParams=params, returnLines=False, winnerTakeAll=wta, parseStyle=arguments.parseStyle) else: # translate to organism before finding most abundant readHits = redistribute.pickBestHitByAbundance( multifile, filterParams=params, returnLines=False, winnerTakeAll=wta, taxonomy=taxonomy, hitStringMap=hitStringMap, parseStyle=arguments.parseStyle) for (read, hit) in readHits: infile_name, read = read.split("/", 1) outhandle = outputMap[unquote_plus(infile_name)] outhandle.write(hit.line.split("/", 1)[1]) if arguments.output_file is not None: for outhandle in outputMap.values(): outhandle.close()
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) 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(): description = __doc__ parser = argparse.ArgumentParser(description=description) add_IO_arguments(parser) add_taxon_arguments(parser, defaults={ 'filter_top_pct': 0, 'parseStyle': ACCS, 'countMethod': 'tophit' }, choices={'countMethod': ('tophit', 'toporg')}) parser.add_argument( "-P", "--proportional", dest="proportional", default=False, action="store_true", help="Assign reads that have multiple equal top hits to taxa such " "that the overal proportion of taxa is consistent with the " "unambiguious hits. This is meant for use with the 'toporg' " "count method.") parser.add_argument( "-i", "--individualFiles", dest="individual", default=False, action="store_true", help="Use this flag to process files independently. Normally, " "counts from all files are pooled for making choices.") add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) # load necessary maps params = FilterParams.create_from_arguments(arguments) if arguments.countMethod == 'toporg': (taxonomy, hitStringMap) = readMaps(arguments) wta = not (arguments.proportional) if len(arguments.input_files) <= 1 or arguments.individual: # loop over input for (inhandle, outhandle) in inputIterator(arguments): logging.debug("Reading from %s and writing to %s" % (inhandle, outhandle)) m8stream = M8Stream(inhandle) if arguments.countMethod == 'tophit': # don't give any taxonomy, just map to accessions for # redistribution readHits = redistribute.pickBestHitByAbundance( m8stream, filterParams=params, returnLines=True, winnerTakeAll=wta, parseStyle=arguments.parseStyle) else: # translate to organism before finding most abundant readHits = redistribute.pickBestHitByAbundance( m8stream, filterParams=params, returnLines=True, winnerTakeAll=wta, taxonomy=taxonomy, hitStringMap=hitStringMap, parseStyle=arguments.parseStyle) for line in readHits: outhandle.write(line) else: # process all files at once multifile = redistribute.multipleFileWrapper(arguments.input_files) # Build a map from input file name to output handle outputMap = {} for infile_handle in arguments.input_files: infile_name = infile_handle.name if arguments.output_file is None: outputMap[infile_name] = sys.stdout elif len(arguments.input_files) <= 1: outputMap[infile_name] = open(arguments.output_file, 'w') else: # use outfileName as suffix if arguments.cwd: # strip path info first (infilePath, infileFile) = os.path.split(infile_name) outfile = "./" + infileFile + arguments.output_file else: outfile = infile_name + arguments.output_file outputMap[infile_name] = open(outfile, 'w') if arguments.countMethod == 'tophit': # don't give any taxonomy, just map to accessions for # redistribution readHits = redistribute.pickBestHitByAbundance( multifile, filterParams=params, returnLines=False, winnerTakeAll=wta, parseStyle=arguments.parseStyle) else: # translate to organism before finding most abundant readHits = redistribute.pickBestHitByAbundance( multifile, filterParams=params, returnLines=False, winnerTakeAll=wta, taxonomy=taxonomy, hitStringMap=hitStringMap, parseStyle=arguments.parseStyle) for (read, hit) in readHits: infile_name, read = read.split("/", 1) outhandle = outputMap[unquote_plus(infile_name)] outhandle.write(hit.line.split("/", 1)[1]) if arguments.output_file is not None: for outhandle in outputMap.values(): outhandle.close()
def main(): description = """ Given two lists of taxids and one or more hit tables, identify reads that: (1) have their best hits in taxid list 1 (2) have all other hits in either list Finally, print out either the hits (that match the target group) for these reads or just read names (-r). The -F filter limits which hits are used in part (2) as well as which are printed. The countMethod (-C) option is not used. """ parser = argparse.ArgumentParser(description=description) add_IO_arguments(parser) add_taxon_arguments( parser, defaults={ 'mapFile': None, 'parseStyle': ACCS, 'filter_top_pct': -1, 'countMethod': 'all', 'taxdir': None}) parser.add_argument( "-g", "--targetTaxonGroup", dest="group1", default=None, metavar="TAXON", action='append', help="Taxon to identify reads in. Top hits (as defined by " "--topHitPct) must be in this group. It can be a taxid, " "a name, or a file listing taxids. Use multiple times to " "specify a list of organisms. Use -a to specify whether " "all or at least one of the top hits must match.") parser.add_argument( "-a", "--any", default=False, action="store_true", help="If specified, accept reads where any top hit is to an organism " "in the target taxon/taxa. By default, all top hits must be " "in the target group.") parser.add_argument( '-t', '--topHitPct', default=0, type=float, help="How close(as a percentage to the best score a hit must be " "to qualify as a top hit. Default is 0, ie must have the best " "score. Use 100 to get all hits.") parser.add_argument( "-G", "--outerTaxonGroup", dest="group2", default=None, metavar="TAXON", action="append", help="Broader taxon to limit reads. All hits (use -F to limit " "these hits) must be in the target group or this group. Again, " "it can be a taxid, a name, or a file listing taxids. " "It can also be inkoved multiple times to choose multiple " "groups.") parser.add_argument( '-r', '--reads', default=False, action="store_true", help="Output just read names. By default, print the relevant hit " "lines for each read") # log level and help add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) # check args if arguments.group1 is None: parser.error("Please use -g to specify a target taxonomic group") if arguments.taxdir is not None: taxonomy = readTaxonomy(arguments.taxdir, namesMap=True) else: taxonomy = None group_1_set = get_group_set(arguments.group1, taxonomy) group_2_set = get_group_set(arguments.group2, taxonomy) logging.debug( "Group 1 has %d entries and 439482 in group1 is %s" % (len(group_1_set), 439482 in group_1_set)) if group_2_set is not None: logging.debug( "Group 2 has %d entries and 439482 in group2 is %s" % (len(group_2_set), 439482 in group_2_set)) # map reads to hits if arguments.parseStyle == GIS: keyType = int else: keyType = None accToTaxMap = parseMapFile( arguments.mapFile, valueType=int, keyType=keyType) # set up some function pointers global hitRE hitRE = parsingREs.get(arguments.parseStyle, None) if arguments.parseStyle == ORGS: getTaxid = _getOrgTaxid elif arguments.parseStyle == HITID: getTaxid = _getHitidTaxid elif arguments.parseStyle == HITDESC: getTaxid = _getHitdescTaxid else: getTaxid = _getExprTaxid # for filtering: filterParams = FilterParams.create_from_arguments(arguments) logging.debug(repr(filterParams)) # loop over hit tables for (inhandle, outhandle) in inputIterator(arguments): readCount = 0 goodReadCount = 0 printCount = 0 # parse file for ( read, hits) in filterM8Stream( inhandle, filterParams, return_lines=False): readCount += 1 bestScore = 0 hitTaxids = {} for hit in hits: score = hit.score taxids = [] # does this hit have at least one associated taxid in group2? for taxid in getTaxid(hit, accToTaxMap, taxonomy): if taxid is None: break if group_2_set is not None and taxid not in group_2_set: break taxids.append(taxid) if len(taxids) == 0: # nothing matched in the wider group break hitTaxids[hit] = taxids # find the top score if score > bestScore: bestScore = score else: # if we get here, then every hit was in wider taxon list logging.debug( "Checking best hits for %s (top score: %.1f)" % (read, bestScore)) all = True recognized = [] for hit, taxids in _getBestHitTaxids( hitTaxids, bestScore, arguments.topHitPct): if _anyTaxidInGroup(taxids, group_1_set): logging.debug("%s (%r) is in group 1" % (hit, taxids)) recognized.append(hit) else: logging.debug( "%s (%r) is not in group 1" % (hit, taxids)) all = False if len(recognized) == 0: # if none of the best are in our target list, next read logging.debug( "No best hits for %s are in group 1" % (read)) continue if (not arguments.any) and (not all): # next read unless user said any or all hits are in list logging.debug( "Not all best hits for %s are in group 1" % (read)) continue # if we get here, then the read is a match goodReadCount += 1 if arguments.reads: logging.debug("Keeping %s" % (read)) outhandle.write(read) outhandle.write('\n') else: logging.debug( "Keeping %d hits for %s" % (len(recognized), read)) for hit in sorted( recognized, key=lambda h: ( h.score, h.hit)): outhandle.write(hit.getLine(filterParams)) printCount += 1 if arguments.reads: logging.info("Printed %d of %d reads" % (goodReadCount, readCount)) else: logging.info( "Printed %d lines for %d of %d reads" % (printCount, goodReadCount, readCount))
def main(): description = """ Given two lists of taxids and one or more hit tables, identify reads that: (1) have their best hits in taxid list 1 (2) have all other hits in either list Finally, print out either the hits (that match the target group) for these reads or just read names (-r). The -F filter limits which hits are used in part (2) as well as which are printed. The countMethod (-C) option is not used. """ parser = argparse.ArgumentParser(description=description) add_IO_arguments(parser) add_taxon_arguments( parser, defaults={"mapFile": None, "parseStyle": ACCS, "filterPct": -1, "countMethod": "all", "taxdir": None} ) parser.add_argument( "-g", "--targetTaxonGroup", dest="group1", default=None, metavar="TAXON", action="append", help="Taxon to identify reads in. Top hits (as defined by " "--topHitPct) must be in this group. It can be a taxid, " "a name, or a file listing taxids. Use multiple times to " "specify a list of organisms. Use -a to specify whether " "all or at least one of the top hits must match.", ) parser.add_argument( "-a", "--any", default=False, action="store_true", help="If specified, accept reads where any top hit is to an organism " "in the target taxon/taxa. By default, all top hits must be " "in the target group.", ) parser.add_argument( "-t", "--topHitPct", default=0, type=float, help="How close(as a percentage to the best score a hit must be " "to qualify as a top hit. Default is 0, ie must have the best " "score. Use 100 to get all hits.", ) parser.add_argument( "-G", "--outerTaxonGroup", dest="group2", default=None, metavar="TAXON", action="append", help="Broader taxon to limit reads. All hits (use -F to limit " "these hits) must be in the target group or this group. Again, " "it can be a taxid, a name, or a file listing taxids. " "It can also be inkoved multiple times to choose multiple " "groups.", ) parser.add_argument( "-r", "--reads", default=False, action="store_true", help="Output just read names. By default, print the relevant hit " "lines for each read", ) # log level and help add_universal_arguments(parser) arguments = parser.parse_args() setup_logging(arguments) # check args if arguments.group1 is None: parser.error("Please use -g to specify a target taxonomic group") if arguments.taxdir is not None: taxonomy = readTaxonomy(arguments.taxdir, namesMap=True) else: taxonomy = None group_1_set = get_group_set(arguments.group1, taxonomy) group_2_set = get_group_set(arguments.group2, taxonomy) logging.debug("Group 1 has %d entries and 439482 in group1 is %s" % (len(group_1_set), 439482 in group_1_set)) if group_2_set is not None: logging.debug("Group 2 has %d entries and 439482 in group2 is %s" % (len(group_2_set), 439482 in group_2_set)) # map reads to hits if arguments.parseStyle == GIS: keyType = int else: keyType = None accToTaxMap = parseMapFile(arguments.mapFile, valueType=int, keyType=keyType) # set up some function pointers global hitRE hitRE = parsingREs.get(arguments.parseStyle, None) if arguments.parseStyle == ORGS: getTaxid = _getOrgTaxid elif arguments.parseStyle == HITID: getTaxid = _getHitidTaxid elif arguments.parseStyle == HITDESC: getTaxid = _getHitdescTaxid else: getTaxid = _getExprTaxid # for filtering: filterParams = FilterParams.create_from_arguments(arguments) logging.debug(repr(filterParams)) # loop over hit tables for (inhandle, outhandle) in inputIterator(arguments): readCount = 0 goodReadCount = 0 printCount = 0 # parse file for (read, hits) in filterM8Stream(inhandle, filterParams, returnLines=False): readCount += 1 bestScore = 0 hitTaxids = {} for hit in hits: score = hit.score taxids = [] # does this hit have at least one associated taxid in group2? for taxid in getTaxid(hit, accToTaxMap, taxonomy): if taxid is None: break if group_2_set is not None and taxid not in group_2_set: break taxids.append(taxid) if len(taxids) == 0: # nothing matched in the wider group break hitTaxids[hit] = taxids # find the top score if score > bestScore: bestScore = score else: # if we get here, then every hit was in wider taxon list logging.debug("Checking best hits for %s (top score: %.1f)" % (read, bestScore)) all = True recognized = [] for hit, taxids in _getBestHitTaxids(hitTaxids, bestScore, arguments.topHitPct): if _anyTaxidInGroup(taxids, group_1_set): logging.debug("%s (%r) is in group 1" % (hit, taxids)) recognized.append(hit) else: logging.debug("%s (%r) is not in group 1" % (hit, taxids)) all = False if len(recognized) == 0: # if none of the best are in our target list, next read logging.debug("No best hits for %s are in group 1" % (read)) continue if (not arguments.any) and (not all): # next read unless user said any or all hits are in list logging.debug("Not all best hits for %s are in group 1" % (read)) continue # if we get here, then the read is a match goodReadCount += 1 if arguments.reads: logging.debug("Keeping %s" % (read)) outhandle.write(read) outhandle.write("\n") else: logging.debug("Keeping %d hits for %s" % (len(recognized), read)) for hit in sorted(recognized, key=lambda h: (h.score, h.hit)): outhandle.write(hit.getLine(filterParams)) printCount += 1 if arguments.reads: logging.info("Printed %d of %d reads" % (goodReadCount, readCount)) else: logging.info("Printed %d lines for %d of %d reads" % (printCount, goodReadCount, readCount))