def main(argv):
args = vars(parser.parse_args())
# read the input fasta_files to be mapped to create read2origin map
fh = open(args["input_mapping"], "r")
lines = fh.readlines()
fh.close()
# create preferred megan mapping
ncbi_megan_map = {}
with open(args["ncbi_megan_map"], 'r') as meganfile:
for line in meganfile:
fields = line.split("\t")
fields = map(str.strip, fields)
ncbi_megan_map[fields[0]] = fields[1]
read2origin = {} # hash mapping input sequences to their original header
# fill read2origin
for l in lines:
fields =l.split("\t")
fields = map(str.strip, fields)
read_name = fields[0]
gi_line = fields[1].strip("\n")
gi_line = gi_line.split("|")
ncbi_id = re.sub("\.[0-9]+$", "", gi_line[3] )
read2origin[fields[0]] = ncbi_id
# Parse NCBI: summary table
fh = open(args["sum_table"], "r")
lines = fh.readlines()
fh.close()
ncbiID_to_taxaID = {} # ncbiID to NCBI Tree taxaID
taxaID_to_taxa = {} # ncbiID to full taxa name
for l in lines:
fields = l.split("\t")
fields = map(str.strip, fields)
ncbi_id = re.sub("\.[0-9]+$", "", fields[0])
tax_id = fields[3]
tax = fields[5]
if ncbi_id not in ncbiID_to_taxaID:
ncbiID_to_taxaID[ncbi_id] = tax_id
if tax_id not in taxaID_to_taxa:
taxaID_to_taxa[tax_id] = tax
# read functional and taxonomic table
fh = open(args["ft_table"], "r")
header = fh.readline()
header = header.split("\t") # list of headers
lines = fh.readlines()
contig_to_orfs = {} # get a list of ORFs for a specific contig
orfs_to_lca = {} # get the lca taxonomy
for l in lines:
fields = l.split("\t")
# ORF_ID ORF_length start end Contig_Name Contig_length strand ec taxonomy product
orf_id = fields[0]
contig = fields[4]
lca = fields[8]
if contig not in contig_to_orfs:
contig_to_orfs[contig] = []
contig_to_orfs[contig].append(orf_id)
if orf_id not in orfs_to_lca:
orfs_to_lca[orf_id] = lca
# get original taxonomy
# print taxaID_to_taxa[ncbiID_to_taxaID[read2origin[contig]]]
# Build the LCA Star NCBI Tree
print "Loading LCAStar:"
lcastar = LCAStar(args["ncbi_tree"])
print "Done."
# set LCAStar parameters
lcastar.setLCAStarParameters(min_depth = 1, alpha = 0.5, min_reads = 1 )
# small helper to translate the list
def get_orfs_taxa(orfs):
list = []
for o in orfs:
list.append(orfs_to_lca[o])
return list
output = open(args["output"], "w")
header = "\t".join(["contig","real","taxa","method","dist", "wtd", "real_linage"])
output.write(header + "\n")
for c in contig_to_orfs:
# the read taxa
contig = c
real = taxaID_to_taxa[ncbiID_to_taxaID[read2origin[c]]]
taxa_list = get_orfs_taxa(contig_to_orfs[c])
lca_list = []
for t in taxa_list:
lca_list.append([ t ])
taxon = lcastar.lca_star(taxa_list)
real_lineage = lcastar.get_lineage(lcastar.get_a_Valid_ID([real]))
lineage = []
for id in real_lineage:
lineage.append(translate_to_prefered_name(id, ncbi_megan_map, lcastar))
real_lineage = lineage
line = "\t".join([contig, real, translate_to_prefered_name(lcastar.get_a_Valid_ID([taxon]),ncbi_megan_map, lcastar), "LCA_Star", str(lcastar.get_distance(taxon, real)), str(lcastar.wtd_distance(real, taxon)), ";".join(real_lineage[::-1])])
output.write(line + "\n")
taxon = lcastar.lca_majority(taxa_list)
line = "\t".join([contig, real, translate_to_prefered_name(lcastar.get_a_Valid_ID([taxon]),ncbi_megan_map, lcastar), "Majority", str(lcastar.get_distance(taxon, real)), str(lcastar.wtd_distance(real, taxon)), ";".join(real_lineage[::-1])])
output.write(line + "\n")
taxon = lcastar.getTaxonomy(lca_list)
line = "\t".join([contig, real, translate_to_prefered_name(lcastar.get_a_Valid_ID([taxon]),ncbi_megan_map, lcastar), "LCA_Squared", str(lcastar.get_distance(taxon, real)), str(lcastar.wtd_distance(real, taxon)), ";".join(real_lineage[::-1])])
output.write(line + "\n")
output.close()
def main(argv):
# parse arguments
args = vars(parser.parse_args())
## read input mapping file (.mapping.txt) to create read2origin map
contig2origin = {
} # dictionary/hash table mapping input sequences to their original header, e.g. >[Header]
contig2origin = create_contig2origin(args["mapping_file"])
## create preferred mapping
ncbi_megan_map = {
} # hash map from given taxonomy to preferred one used by megan
with open(args["ncbi_megan_map"], 'r') as meganfile:
for line in meganfile:
fields = line.split("\t")
fields = map(str.strip, fields)
ncbi_megan_map[fields[0]] = fields[1]
# contig_taxa_data structure
contig_to_taxa = {}
# Read blast table
with open(args["parsed_blast"], "r") as fh:
for l in fh:
if re.match("^\#", l):
clean_tab_lines(l)
else:
fields = clean_tab_lines(l)
contig_hits = contig_pattern.search(fields[0])
if contig_hits:
contig = contig_hits.group(1)
orf = contig_hits.group(2)
# add to data structre if it doesn't exist
if contig not in contig_to_taxa:
contig_to_taxa[contig] = {}
if orf not in contig_to_taxa[contig]:
contig_to_taxa[contig][orf] = []
# pull taxonomy out of annotation
taxa_hits = taxonomy_pattern.search(fields[9])
if taxa_hits:
taxa = taxa_hits.group(1)
bitscore = fields[3]
contig_to_taxa[contig][orf].append(
(taxa, float(str(bitscore))))
else:
continue
else:
continue
## Load contig references (if available or applicable)
# read contig taxa reference if available
contig_to_taxa_ref = None
if args["contig_taxa_ref"]:
contig_to_taxa_ref = {}
if args["contig_taxa_ref"]:
with open(args["contig_taxa_ref"], "r") as fh:
for l in fh:
fields = clean_tab_lines(l)
contig_id = fields[0]
contig_origin = fields[1]
contig_to_taxa_ref[contig_id] = contig_origin
# all contigs hypothetically have the same reference origin (i.e., single cells)
sample_ref = None
if args["sample_taxa_ref"]:
sample_ref = args["sample_taxa_ref"]
## Build the LCA Star NCBI Tree
lcastar = LCAStar(args["ncbi_tree"])
lcastar.setLCAStarParameters(min_depth=1, alpha=args["alpha"], min_reads=1)
## Calculate LCA for each ORF
contig_to_lca = {}
for contig in contig_to_taxa:
for orf in contig_to_taxa[contig]:
if contig not in contig_to_lca:
contig_to_lca[contig] = {}
if orf not in contig_to_lca[contig]:
contig_to_lca[contig][orf] = None
contig_taxas = contig_to_taxa[contig][orf]
if len(contig_taxas) == 0:
contig_to_lca[contig][orf] = "root"
else:
if args['orf_summary'] == 'besthit':
contig_taxas.sort(key=operator.itemgetter(1), reverse=True)
best_blast_taxa = contig_taxas[0][0]
contig_to_lca[contig][orf] = best_blast_taxa
elif args['orf_summary'] == 'orf_majority':
majority_list = []
for t in contig_taxas:
majority_list.append(t[0])
#TODO Update to check for alterantive taxonomy names
contig_to_lca[contig][orf] = lcastar.simple_majority(
majority_list)
else:
lca_list = [] # create a list of lists for LCA calculation
for t in contig_taxas:
lca_list.append([t[0]])
contig_to_lca[contig][orf] = lcastar.getTaxonomy(lca_list)
## calculate taxonomy statistics LCA, for each ORF
# contig_to_taxa = {}
## LCA^2, Majority, and LCA* for each ORF
writeout(args, contig_to_lca, contig_to_taxa_ref, sample_ref, lcastar,
ncbi_megan_map)
def main(argv):
# parse arguments
args = vars(parser.parse_args())
## read input mapping file (.mapping.txt) to create read2origin map
contig2origin = {} # dictionary/hash table mapping input sequences to their original header, e.g. >[Header]
contig2origin = create_contig2origin(args["mapping_file"])
## create preferred mapping
ncbi_megan_map = {} # hash map from given taxonomy to preferred one used by megan
with open(args["ncbi_megan_map"], 'r') as meganfile:
for line in meganfile:
fields = line.split("\t")
fields = map(str.strip, fields)
ncbi_megan_map[fields[0]] = fields[1]
# contig_taxa_data structure
contig_to_taxa = {}
# Read blast table
with open(args["parsed_blast"], "r") as fh:
for l in fh:
if re.match("^\#", l):
clean_tab_lines(l)
else:
fields = clean_tab_lines(l)
contig_hits = contig_pattern.search(fields[0])
if contig_hits:
contig = contig_hits.group(1)
orf = contig_hits.group(2)
# add to data structre if it doesn't exist
if contig not in contig_to_taxa:
contig_to_taxa[contig] = {}
if orf not in contig_to_taxa[contig]:
contig_to_taxa[contig][orf] = []
# pull taxonomy out of annotation
taxa_hits = taxonomy_pattern.search(fields[9])
if taxa_hits:
taxa = taxa_hits.group(1)
bitscore = fields[3]
contig_to_taxa[contig][orf].append( (taxa, float(str(bitscore))) )
else:
continue
else:
continue
## Load contig references (if available or applicable)
# read contig taxa reference if available
contig_to_taxa_ref = None
if args["contig_taxa_ref"]:
contig_to_taxa_ref = {}
if args["contig_taxa_ref"]:
with open(args["contig_taxa_ref"], "r") as fh:
for l in fh:
fields = clean_tab_lines(l)
contig_id = fields[0]
contig_origin = fields[1]
contig_to_taxa_ref[contig_id] = contig_origin
# all contigs hypothetically have the same reference origin (i.e., single cells)
sample_ref = None
if args["sample_taxa_ref"]:
sample_ref = args["sample_taxa_ref"]
## Build the LCA Star NCBI Tree
lcastar = LCAStar(args["ncbi_tree"])
lcastar.setLCAStarParameters(min_depth = 1, alpha = args["alpha"], min_reads = 1)
## Calculate LCA for each ORF
contig_to_lca = {}
for contig in contig_to_taxa:
for orf in contig_to_taxa[contig]:
if contig not in contig_to_lca:
contig_to_lca[contig] = {}
if orf not in contig_to_lca[contig]:
contig_to_lca[contig][orf] = None
contig_taxas = contig_to_taxa[contig][orf]
if len(contig_taxas) == 0:
contig_to_lca[contig][orf] = "root"
else:
if args['orf_summary'] == 'besthit':
contig_taxas.sort(key=operator.itemgetter(1), reverse=True)
best_blast_taxa = contig_taxas[0][0]
contig_to_lca[contig][orf] = best_blast_taxa
elif args['orf_summary'] == 'orf_majority':
majority_list = []
for t in contig_taxas:
majority_list.append(t[0])
#TODO Update to check for alterantive taxonomy names
contig_to_lca[contig][orf] = lcastar.simple_majority(majority_list)
else:
lca_list = [] # create a list of lists for LCA calculation
for t in contig_taxas:
lca_list.append([t[0]])
contig_to_lca[contig][orf] = lcastar.getTaxonomy(lca_list)
## calculate taxonomy statistics LCA, for each ORF
# contig_to_taxa = {}
## LCA^2, Majority, and LCA* for each ORF
writeout(args, contig_to_lca, contig_to_taxa_ref, sample_ref, lcastar, ncbi_megan_map)
def main():
args = vars(parser.parse_args())
# parse options
parsed_blast = args["parsed_blast"][0]
pwy_long = args["pwy_long"][0]
ncbi_tree = args["ncbi_tree"][0]
megan_names = args["megan_names"][0]
output_file = args["output_file"][0]
lca = args["lca"]
rpkm_file = args["rpkm"][0]
# parse RPKM value for every ORF
if rpkm_file:
orf_id_to_rpkm = {}
orf_id_to_rpkm = process_rpkm_file(rpkm_file, orf_id_to_rpkm)
# Load NCBI Tree and LCA Star object
global lcastar
lcastar = LCAStar(ncbi_tree, megan_names)
# LCA star parameters
alpha = 0.51
min_reads = 0
lcastar.setLCAStarParameters(0, alpha, min_reads)
print 'Done initializing NCBI Tree'
# capture taxonomic annotation from parsed B/LAST file
blast_to_taxonomy = {} # map from read to taxonomy
blast_to_taxonomy = read_and_clean_parsed_blast(parsed_blast,
blast_to_taxonomy, lca)
# read through pathway file and append taxonomy field if hit occurs:
header = [
"SAMPLE", "PWY_NAME", "PWY_COMMON_NAME", "RXN_NAME", "RXN_COMMON_NAME",
"NUM_REACTIONS", "NUM_COVERED_REACTIONS", "ORF_COUNT", "ORF",
"TAXONOMY"
]
if orf_id_to_rpkm:
header.append("RPKM")
if ".gz" in pwy_long:
with gzip.open(output_file + ".gz", 'w') as fh_out:
fh_out.write("\t".join(header) + "\n")
with gzip.open(pwy_long, 'rb') as fh:
fh.readline() # read out old header
for line in fh:
fields = line.split("\t")
orf_id = fields[-1].strip("\n")
fields[-1] = fields[-1].strip("\n")
if orf_id in blast_to_taxonomy:
fields.append(blast_to_taxonomy[orf_id])
else:
fields.append("NA")
if orf_id_to_rpkm:
if orf_id in orf_id_to_rpkm:
fields.append(orf_id_to_rpkm[orf_id])
else:
fields.append("NA")
fh_out.write("\t".join(fields) + "\n")
else:
with open(output_file + ".gz", 'wb') as fh_out:
fh_out.write("\t".join(header) + "\n")
with open(pwy_long, 'rb') as fh:
fh.readline() # read out old header
for line in fh:
fields = line.split("\t")
orf_id = fields[-1].strip("\n")
fields[-1] = fields[-1].strip("\n")
if orf_id in blast_to_taxonomy:
fields.append(blast_to_taxonomy[orf_id])
else:
fields.append("NA")
fh_out.write("\t".join(fields) + "\n")
exit()
