def direct_match(db_file_name,patterns,max_errors):
try:
num_patterns = len(patterns)
except TypeError:
error = "Error: Invalid type of pattern data"
raise Exception(error)
if type(max_errors) is type([1]):
for elm in max_errors:
if elm>8:
print "An element of max errors is too high, changing it to 8"
elm=8
if elm<1:
print "An element of max errors is too low, changing it to 1"
elm=1
else:
if type(max_errors) is type(1):
if max_errors>8:
print "The maximum error rate is to high, changing it to 8"
max_errors=8
if max_errors<1:
print "The maximum error rate is too low, changing it to 1"
max_errors=1
errors = max_errors
max_errors = []
for i in range(num_patterns):
max_errors.append(errors)
else:
error = "Error: Invalid type for number of errors"
raise Exception(error)
#check at least one query has been specified
if num_patterns<1:
error = "Error: At least one pattern to find must be specified"
raise Exception(error)
matchers = []
#deal with a single query specified as a string rather than a list
patterns_to_match = []
if type(patterns)==type("abc"):
patterns_to_match.append(patterns)
num_patterns=1
else:
patterns_to_match=patterns
#create a pattern object for each query
i=0
for pattern in patterns_to_match:
#if the pattern is too long simply abort, we don't want to deal
#with overflow issues and there are better tools out there anyway
#similarly if it less than 2 aa then dont proceed
if not 1<len(pattern)<60:
error = "Error: A query protein had a length outside the range supported by this program"
raise Exception(error)
matchers.append(compile(pattern,len(pattern),max_errors[i]))
i+=1
#open the specified database for reading
try:
db_file = open(db_file_name, 'r')
except IOError:
error = "Error: Cannot open specified db file"
raise Exception(error)
#storage for database information
rec = []
header = ''
#storage for information about each match
#this allows for later calculation of p values and
#output of this information to the user
results_record_headers = []
results_surrounding_text = []
num_match_recs = 0
#begin parsing database
for line in db_file:
#database body section
if line[0] != '>':
rec.append(line[:-1])
#database header
else:
#attempt to match with previous body section
record = ''.join(rec)
begin_rec = 0
for i in range(num_patterns):
matches = agrepy(patterns_to_match[i], len(patterns_to_match[i]),\
record,len(record),1,matchers[i])
if matches:
if begin_rec==0:
begin_rec = 1
num_match_recs +=1
results_record_headers.append(split(header)[1])
pattern_matches = []
for elm in matches:
#for some large strings on 64 bit platforms sagrepy can return invalid
#match data, so remove this here rather than risk problems later
if elm[0]<elm[1]:
pattern_matches.append(record[elm[0]:elm[1]])
results_surrounding_text.append(pattern_matches)
rec = []
header = line
#attempt to match the final body section
record = ''.join(rec)
begin_rec = 0
for i in range(num_patterns):
matches = agrepy(patterns_to_match[i],\
len(patterns_to_match[i]),record,len(record),1,matchers[i])
if matches:
if begin_rec==0:
begin_rec = 1
num_match_recs +=1
results_record_headers.append(split(header)[1])
pattern_matches = []
for elm in matches:
#for some large strings on 64 bit platforms sagrepy can return invalid
#match data, so remove this here rather than risk problems later
if elm[0]<elm[1]:
pattern_matches.append(record[elm[0]:elm[1]])
results_surrounding_text.append(pattern_matches)
#end parsing database
#now calculate p values and output results to the user
outputResults(patterns_to_match,num_match_recs,results_record_headers,results_surrounding_text)
db_file.close()
#Finally return result headers to the caller in case
#they wish to make use of them in some other way
return results_record_headers
def match(self,patterns,max_errors):
""" Parameters:
patterns - a string (i.e. "MKFL")
- a single list element (i.e. ["MKFL"])
- multiple list elements (i.e. ["LIL","MKFL","CLF"])
- grouped list elements (i.e. [["LIL","MKFL"],"CLF"])
The first two cases are equivalent, and will result in
a search for all records that contain the string/element
Multiple list elements will result in a search for any
records that contain ANY of the elements in the list
Elements in a list can also be grouped together. If a
group contains only a single element it will not have
any effect, but specifying more than one element in a
group has the effect that only records that contain ALL
elements in a group will be searched for (but if there
is more than one group a record needs to contain only one
group to match, in other words elements within a group
are ANDed together, and the groups are ORed). Also it
is possible that false drops will return some subset
of the patterns even if they are not explicitly
searched for. The main advantage of grouping elements
together like this is dramatically faster searches.
max-errors - an integer
- a list (i.e [2,3,1])
A single integer value will be applied to ALL patterns.
If a seperate error rate is desired for each pattern then
specify a list with 1 integer per pattern
"""
#---------------------------
#1. Check validity of inputs
#---------------------------
#put patterns into uniform list format
#first deal with a single query specified as a string rather than a list
tmp_patterns = []
if type(patterns)==type("abc"):
tmp_patterns.append(patterns)
else:
tmp_patterns=patterns
#now pull out all groups, storing the info of each
groups = []
patterns_to_match = []
i=0
for pat in tmp_patterns:
if type(pat)==type([]):
group = []
for elm in pat:
patterns_to_match.append(elm)
group.append(i)
i+=1
groups.append(group)
else:
patterns_to_match.append(pat)
i+=1
#add an int for patterns without groups
if len(groups)==0:
groups.append([-1])
#Finally get some data about this list
try:
num_patterns = len(patterns_to_match)
except TypeError:
error = "Error: Invalid type of pattern data"
raise Exception(error)
#and check at least one query has been specified
if num_patterns<1:
error = "Error: At least one pattern to find must be specified"
raise Exception(error)
#Now move on to checking max_errors
if type(max_errors) is type([]):
i=0
for elm in max_errors:
if elm>8:
print "An element of max errors is too high, changing it to 8"
elm=8
if elm<1:
print "An element of max errors is too low, changing it to 1"
elm=1
i+=1
if i>num_patterns:
error = "Error: More errors were specified than patterns given"
raise Exception(error)
if i<num_patterns:
error = "Error: Less errors were specified than patterns given"
raise Exception(error)
else:
if type(max_errors) is type(1):
if max_errors>8:
print "The maximum error rate is to high, changing it to 8"
max_errors=8
if max_errors<1:
print "The maximum error rate is too low, changing it to 1"
max_errors=1
errors = max_errors
max_errors = []
for i in range(num_patterns):
max_errors.append(errors)
else:
error = "Error: Invalid type for number of errors"
print error
return error
#------------------------
#2. Encode Query Patterns
#------------------------
matchers = []
query_pieces = []
piece_sizes,segments = self.PIECE_SIZES,self.SEGMENTS
i=0
for pattern in patterns_to_match:
#create a pattern object for each query
matchers.append(compile(pattern,len(pattern),max_errors[i]))
#check the query length is within acceptable bounds
pattern_length = len(pattern)
#if the pattern is too long simply abort, we don't want to deal
#with overflow issues and there are better tools out there anyway
if pattern_length>self.MAX_QUERY_LENGTH:
error = []
error.append("Error: This program currently cannot deal with ")
error.append("proteins longer than ")
error.append(str(self.MAX_QUERY_LENGTH))
error.append(" aa, sorry")
error = ''.join(error)
clearMatchInfo()
raise Exception(error)
#break the query into a number of segments equal to the
#maximum number of errors for the query plus one
num_segs = max_errors[i]+1
segs_size=pattern_length/num_segs
#if the pattern is too short then abort all
#preprocessed matching and give all input to direct
#match to deal with and return to the user
if segs_size<piece_sizes[0]:
print "A query had too many errors for its size"
print "Using direct_match instead..."
clearMatchInfo()
return direct_match(self.database_name,patterns,max_errors)
#for each segment encode all pieces
createSegments(num_segs)
completed=0
for j in range(num_segs):
#on the last one make size the leftover
if j+1==num_segs:
segs_size = pattern_length-(segs_size*(num_segs-1))
for elm in segments[segs_size]:
encodeQueryPiece(j,pattern[completed:completed+elm],elm-2)
completed+=elm
for elm in groups:
if i in elm:
updateDropsetAnd()
else:
updateDropsetOr()
i+=1
#-------------------------
#3. Set up database access
#-------------------------
try:
db_file = open(self.database_name, 'r')
except IOError:
error = "Error: Cannot open specified db file"
raise Exception(error)
#----------------------------------------------
#4. Search dropped database records for matches
#----------------------------------------------
#storage for information about each match
#this allows for later calculation of p values and
#output of this information to the user
results_record_headers = []
results_surrounding_text = []
num_match_recs = 0
#handles for called functions
index = self.header_index
db_pos = db_file.seek
#start parsing dropfile
rec_num=nextDrop(1)
while(rec_num):
db_pos(index[rec_num-1])
title = db_file.readline().split()[1]
rec = []
line = db_file.readline()
while(line and line[0]!='>'):
rec.append(line[:-1])
line = db_file.readline()
record = ''.join(rec)
begin_rec = 0
for i in range(num_patterns):
matches = agrepy(patterns_to_match[i], len(patterns_to_match[i]),\
record,len(record),1,matchers[i])
if matches:
if begin_rec==0:
begin_rec = 1
num_match_recs +=1
results_record_headers.append(title)
pattern_matches = []
for elm in matches:
start=0
if elm[0]>0:
start = elm[0]
pattern_matches.append(record[start:elm[1]])
results_surrounding_text.append(pattern_matches)
rec_num=nextDrop(rec_num+1)
clearMatchInfo()
db_file.close()
#--------------------
#5. Deal with results
#--------------------
outputResults(patterns_to_match,num_match_recs,results_record_headers,
results_surrounding_text)
#Finally return result headers to the caller in case
#they wish to make use of them in some other way
return results_record_headers
