def find_schemas(self, fitness, num_schemas):
"""Find the given number of unique schemas using a genetic algorithm
Arguments:
o fitness - A callable object (ie. function) which will evaluate
the fitness of a motif.
o num_schemas - The number of unique schemas with good fitness
that we want to generate.
"""
start_population = \
Organism.function_population(self.motif_generator.random_motif,
self.initial_population,
fitness)
finisher = SimpleFinisher(num_schemas, self.min_generations)
# set up the evolver and do the evolution
evolver = GenerationEvolver(start_population, self.selector)
evolved_pop = evolver.evolve(finisher.is_finished)
# convert the evolved population into a PatternRepository
schema_info = {}
for org in evolved_pop:
# convert the Genome from a MutableSeq to a Seq so that
# the schemas are just strings (and not array("c")s)
seq_genome = org.genome.toseq()
schema_info[seq_genome.data] = org.fitness
return PatternRepository(schema_info)
def from_motifs(self, motif_repository, motif_percent, num_ambiguous):
"""Generate schema from a list of motifs.
Arguments:
o motif_repository - A MotifRepository class that has all of the
motifs we want to convert to Schema.
o motif_percent - The percentage of motifs in the motif bank which
should be matches. We'll try to create schema that match this
percentage of motifs.
o num_ambiguous - The number of ambiguous characters to include
in each schema. The positions of these ambiguous characters will
be randomly selected.
"""
# get all of the motifs we can deal with
all_motifs = motif_repository.get_top_percentage(motif_percent)
# start building up schemas
schema_info = {}
# continue until we've built schema matching the desired percentage
# of motifs
total_count = self._get_num_motifs(motif_repository, all_motifs)
matched_count = 0
assert total_count > 0, "Expected to have motifs to match"
while (float(matched_count) / float(total_count)) < motif_percent:
new_schema, matching_motifs = \
self._get_unique_schema(schema_info.keys(),
all_motifs, num_ambiguous)
# get the number of counts for the new schema and clean up
# the motif list
schema_counts = 0
for motif in matching_motifs:
# get the counts for the motif
schema_counts += motif_repository.count(motif)
# remove the motif from the motif list since it is already
# represented by this schema
all_motifs.remove(motif)
# all the schema info
schema_info[new_schema] = schema_counts
matched_count += schema_counts
# print "percentage:", float(matched_count) / float(total_count)
return PatternRepository(schema_info)
def find(self, seq_records, motif_size):
"""Find all motifs of the given size in the passed SeqRecords.
Arguments:
o seq_records - A list of SeqRecord objects which the motifs
will be found from.
o motif_size - The size of the motifs we want to look for.
Returns:
A PatternRepository object that contains all of the motifs (and their
counts) found in the training sequences).
"""
motif_info = self._get_motif_dict(seq_records, motif_size)
return PatternRepository(motif_info)
def find(self, seq_records, signature_size, max_gap):
"""Find all signatures in a group of sequences.
Arguments:
o seq_records - A list of SeqRecord objects we'll use the sequences
from to find signatures.
o signature_size - The size of each half of a signature (ie. if this
is set at 3, then the signature could be AGC-----GAC)
o max_gap - The maximum gap size between two parts of a signature.
"""
sig_info = self._get_signature_dict(seq_records, signature_size,
max_gap)
return PatternRepository(sig_info)
def find_differences(self, first_records, second_records, motif_size):
"""Find motifs in two sets of records and return the differences.
This is used for finding motifs, but instead of just counting up all
of the motifs in a set of records, this returns the differences
between two listings of seq_records.
o first_records, second_records - Two listings of SeqRecord objects
to have their motifs compared.
o motif_size - The size of the motifs we are looking for.
Returns:
A PatternRepository object that has motifs, but instead of their
raw counts, this has the counts in the first set of records
subtracted from the counts in the second set.
"""
first_motifs = self._get_motif_dict(first_records, motif_size)
second_motifs = self._get_motif_dict(second_records, motif_size)
motif_diffs = {}
# first deal with all of the keys from the first motif
for cur_key in first_motifs.keys():
if second_motifs.has_key(cur_key):
motif_diffs[cur_key] = first_motifs[cur_key] - \
second_motifs[cur_key]
else:
motif_diffs[cur_key] = first_motifs[cur_key]
# now see if there are any keys from the second motif
# that we haven't got yet.
missing_motifs = second_motifs.keys()[:]
# remove all of the motifs we've already added
for added_motif in motif_diffs.keys():
if added_motif in missing_motifs:
missing_motifs.remove(added_motif)
# now put in all of the motifs we didn't get
for cur_key in missing_motifs:
motif_diffs[cur_key] = 0 - second_motifs[cur_key]
return PatternRepository(motif_diffs)
