def process_parsed_blastoutput(dbname, weight, blastoutput, cutoffs, annotation_results):
blastparser = BlastOutputTsvParser(dbname, blastoutput)
fields = ['q_length', 'bitscore', 'bsr', 'expect', 'aln_length', 'identity', 'ec' ]
if cutoffs.taxonomy:
fields.append('taxonomy')
fields.append('product')
annotation = {}
for data in blastparser:
#if count%10000==0:
# print count
if isWithinCutoffs(data, cutoffs) :
#print data['query'] + '\t' + str(data['q_length']) +'\t' + str(data['bitscore']) +'\t' + str(data['expect']) +'\t' + str(data['identity']) + '\t' + str(data['bsr']) + '\t' + data['ec'] + '\t' + data['product']
# if data['query'] =='NapDC_illum_asm_188606_0':
# print dbname
annotation['bsr'] = data['bsr']
annotation['ec'] = data['ec']
annotation['product'] = strip_taxonomy(process_product(data['product'], dbname) )
annotation['value'] = compute_annotation_value(annotation)*weight
# print annotation
if not data['query'] in annotation_results:
annotation_results[data['query']] = {'value':0}
if annotation_results[data['query']]['value'] <= annotation['value'] :
annotation_results[data['query']] = annotation.copy()
# add_refscore_to_file(blastoutput,refscore_file, allNames)
count = len(annotation_results.keys())
return count
def process_parsed_blastoutput(dbname, weight, blastoutput, cutoffs, annotation_results):
blastparser = BlastOutputTsvParser(dbname, blastoutput, shortenorfid=False)
fields = ['q_length', 'bitscore', 'bsr', 'expect', 'aln_length', 'identity', 'ec' ]
if cutoffs.taxonomy:
fields.append('taxonomy')
fields.append('product')
annotation = {}
for data in blastparser:
#if count%10000==0:
if isWithinCutoffs(data, cutoffs) :
#print data['query'] + '\t' + str(data['q_length']) +'\t' + str(data['bitscore']) +'\t' + str(data['expect']) +'\t' + str(data['identity']) + '\t' + str(data['bsr']) + '\t' + data['ec'] + '\t' + data['product']
# if data['query'] =='NapDC_illum_asm_188606_0':
# print dbname
annotation['bsr'] = data['bsr']
annotation['ec'] = data['ec']
annotation['product'] = strip_taxonomy(process_product(data['product'], dbname) )
annotation['value'] = compute_annotation_value(annotation)*weight
# print annotation
if not data['query'] in annotation_results:
annotation_results[data['query']] = {'value':0}
if annotation_results[data['query']]['value'] <= annotation['value'] :
annotation_results[data['query']] = annotation.copy()
# add_refscore_to_file(blastoutput,refscore_file, allNames)
count = len(annotation_results.keys())
return count
def process_product(product, database, similarity_threshold=0.9):
"""Returns the best set of products from the list of (*database*,
*product*) tuples *products*.
Each product in the set is first trimmed down, removing database-specific
information.
The set is then determined by first sorting the products by length
(ascending), and then, for each product, sequentially applying the longest
common substring algorithm to determine the similarity between the product
and already determined products. If this similarity is greater than the
specified *similarity_threshold*, the longer of the two products is chosen
to be a determined product.
"""
processed_product = ''
# print 'dbase', database
# COG
if database == 'cog':
results = re.search(r'Function: (.+?) #', product)
if results:
processed_product = results.group(1)
# KEGG: split and process
elif database == 'kegg':
kegg_products = re.split(r'\s*;\s+', product)
for kegg_product in kegg_products:
# Toss out organism:ID pairs, gene names, and KO IDs
kegg_product = re.sub(r'^lcl[|]', '', kegg_product)
kegg_product = re.sub(r'[a-z]{3}:\S+', '', kegg_product)
kegg_product = kegg_product.strip()
kegg_product = re.sub(r'(, \b[a-z]{3}[A-Z]?\b)+', '', kegg_product)
kegg_product = re.sub(r'^\b[a-z]{3}[A-Z]?\b', '', kegg_product)
# get KO number
kegg_product = re.sub(r'\bK\d{5}\b', '', kegg_product)
# Also toss out anything between square brackets
kegg_product = re.sub(r'\[.*\]', '', kegg_product)
if kegg_product.strip():
processed_product = kegg_product.strip()
# RefSeq: split and process
elif database == 'refseq':
for subproduct in product.split('; '):
subproduct = re.sub(r'[a-z]{2,}\|(.+?)\|\S*', '', subproduct)
subproduct = re.sub(r'\[.+?\]', '', subproduct)
if subproduct.strip():
processed_product = subproduct.strip()
# MetaCyc: split and process
elif database == 'metacyc':
# Pull out first name after the accession code:
product_name = product.split('#')[0].strip()
product_name = re.sub(r'^[^ ]* ', '', product_name)
product_name = re.sub(r' OS=.*', '', product_name)
if product_name:
processed_product = product_name
# Seed: split and process
elif database == 'seed':
for subproduct in product.split('; '):
#subproduct = re.sub(r'[a-z]{2,}\|(.+?)\|\S*', '', subproduct)
subproduct = re.sub(r'\[.+?\]', '', subproduct)
subproduct = re.sub(r'\(.+?\)', '', subproduct)
if subproduct.strip():
processed_product = subproduct.strip()
elif database == 'cazy':
for subproduct in product.split('; '):
#subproduct = re.sub(r'[a-z]{2,}\|(.+?)\|\S*', '', subproduct)
subproduct = re.sub(r'\[.+?\]', '', subproduct)
subproduct = re.sub(r'\(.+?\)', '', subproduct)
if subproduct.strip():
processed_product = subproduct.strip()
print processed_product
# MetaCyc: split and process
# Generic
else:
processed_product = strip_taxonomy(product)
words = [x.strip() for x in processed_product.split()]
filtered_words = []
underscore_pattern = re.compile("_")
arrow_pattern = re.compile(">")
for word in words:
if not underscore_pattern.search(word) and not arrow_pattern.search(
word):
filtered_words.append(word)
#processed_product = ' '.join(filtered_words)
# Chop out hypotheticals
processed_product = remove_repeats(filtered_words)
processed_product = re.sub(';', '', processed_product)
# can actually be a proper annotation
# processed_product = re.sub(r'hypothetical protein','', processed_product)
return processed_product
def process_product(product, database, similarity_threshold=0.9):
"""Returns the best set of products from the list of (*database*,
*product*) tuples *products*.
Each product in the set is first trimmed down, removing database-specific
information.
The set is then determined by first sorting the products by length
(ascending), and then, for each product, sequentially applying the longest
common substring algorithm to determine the similarity between the product
and already determined products. If this similarity is greater than the
specified *similarity_threshold*, the longer of the two products is chosen
to be a determined product.
"""
processed_product = ''
# print 'dbase', database
# COG
if database == 'cog':
results = re.search(r'Function: (.+?) #', product)
if results:
processed_product=results.group(1)
# KEGG: split and process
elif database == 'kegg':
kegg_products = re.split(r'\s*;\s+', product)
for kegg_product in kegg_products:
# Toss out organism:ID pairs, gene names, and KO IDs
kegg_product = re.sub(r'^lcl[|]', '', kegg_product)
kegg_product = re.sub(r'[a-z]{3}:\S+', '', kegg_product)
kegg_product = kegg_product.strip()
kegg_product = re.sub(r'(, \b[a-z]{3}[A-Z]?\b)+', '', kegg_product)
kegg_product = re.sub(r'^\b[a-z]{3}[A-Z]?\b', '', kegg_product)
# get KO number
kegg_product = re.sub(r'\bK\d{5}\b', '', kegg_product)
# Also toss out anything between square brackets
kegg_product = re.sub(r'\[.*\]', '', kegg_product)
if kegg_product.strip():
processed_product=kegg_product.strip()
# RefSeq: split and process
elif database == 'refseq':
for subproduct in product.split('; '):
subproduct = re.sub(r'[a-z]{2,}\|(.+?)\|\S*', '', subproduct)
subproduct = re.sub(r'\[.+?\]', '', subproduct)
if subproduct.strip():
processed_product=subproduct.strip()
# MetaCyc: split and process
elif database == 'metacyc':
# Pull out first name after the accession code:
product_name = product.split('#')[0].strip()
product_name = re.sub(r'^[^ ]* ', '', product_name)
product_name = re.sub(r' OS=.*', '', product_name)
if product_name:
processed_product=product_name
# Seed: split and process
elif database == 'seed':
for subproduct in product.split('; '):
#subproduct = re.sub(r'[a-z]{2,}\|(.+?)\|\S*', '', subproduct)
subproduct = re.sub(r'\[.+?\]', '', subproduct)
subproduct = re.sub(r'\(.+?\)', '', subproduct)
if subproduct.strip():
processed_product=subproduct.strip()
elif database == 'cazy':
for subproduct in product.split('; '):
#subproduct = re.sub(r'[a-z]{2,}\|(.+?)\|\S*', '', subproduct)
subproduct = re.sub(r'\[.+?\]', '', subproduct)
subproduct = re.sub(r'\(.+?\)', '', subproduct)
if subproduct.strip():
processed_product=subproduct.strip()
print processed_product
# MetaCyc: split and process
# Generic
else:
processed_product=strip_taxonomy(product)
words = [ x.strip() for x in processed_product.split() ]
filtered_words =[]
underscore_pattern = re.compile("_")
arrow_pattern = re.compile(">")
for word in words:
if not underscore_pattern.search(word) and not arrow_pattern.search(word):
filtered_words.append(word)
#processed_product = ' '.join(filtered_words)
# Chop out hypotheticals
processed_product = remove_repeats(filtered_words)
processed_product = re.sub(';','',processed_product)
# can actually be a proper annotation
# processed_product = re.sub(r'hypothetical protein','', processed_product)
return processed_product