def test_loading(self):
cluster_string = """cluster,sequence
cluster1,seq1
cluster1,seq2
cluster2,seq3
cluster2,seq4
"""
cluster_map = {'cluster1': ['seq1', 'seq2'], 'cluster2': ['seq3', 'seq4']}
loaded_cluster_map = load_cluster_map(helpers.fake_file(cluster_string), cluster_col='cluster')
self.assertEqual(loaded_cluster_map, cluster_map)
def test_loading(self):
cluster_string = """cluster,sequence
cluster1,seq1
cluster1,seq2
cluster2,seq3
cluster2,seq4
"""
cluster_map = {
'cluster1': ['seq1', 'seq2'],
'cluster2': ['seq3', 'seq4']
}
loaded_cluster_map = load_cluster_map(
helpers.fake_file(cluster_string), cluster_col='cluster')
self.assertEqual(loaded_cluster_map, cluster_map)
def analyze(args):
import logging
logging.captureWarnings(True)
# Fetch sequence records and analysis patterns
seq_records = SeqIO.to_dict(SeqIO.parse(args.alignment, 'fasta'))
patterns = [mut_pattern.patterns[p] for p in args.patterns]
pattern_names = [p.name for p in patterns]
prefix = path.join(args.out_dir, args.prefix)
analysis_settings = dict(rpr_cutoff=args.rpr_cutoff,
significance_level=args.significance_level,
quants=args.quants,
pos_quants_only=args.pos_quants_only,
caller=args.caller,
prior=args.prior,
cdfs=args.cdfs,
quadr_maxiter=args.quadr_maxiter,
optim_maxiter=args.optim_maxiter)
# Need to think about how best to fork things here; for instance, might make sense to let the user specify
# the initial clusters for whatever reason... However, specifying the reference sequences shouldn't make
# any sense there
if args.reference_sequences:
reference_sequences = SeqIO.to_dict(
SeqIO.parse(args.reference_sequences, 'fasta'))
else:
reference_sequences = None
# This lets the cluster map be optional, so that this script can be used
# for naive hm filtering/analysis
cluster_map = load_cluster_map(
args.cluster_map,
cluster_col=args.cluster_col) if args.cluster_map else None
alignments = AlignmentSet(seq_records,
cluster_map,
consensus_threshold=args.consensus_threshold,
reference_sequences=reference_sequences)
# Create the analysis generator
analysis = alignments.multiple_context_analysis(patterns,
**analysis_settings)
if args.cluster_threshold:
for hm_it in range(args.cluster_iterations - 1):
print " ..On hm/cluster iteration", hm_it
# Grab the HM columns from the most recent analysis and split out the pos sites
hm_columns = []
for result in analysis:
hm_columns += result['call']['mut_columns']
hm_neg_aln = Alignment(
seq_records.values()).split_hypermuts(hm_columns).hm_neg_aln
# Cluster with the specified settings
clustering = alnclst.Clustering(hm_neg_aln, args.cluster_threshold,
args.consensus_threshold)
clustering = clustering.recenter(args.recentering_iterations)
clustering.merge_small_clusters(args.min_per_cluster)
cluster_map = parse_clusters(clustering.mapping_iterator(),
cluster_key=0,
sequence_key=1)
# Create the Alignment set
clustered_alignment = AlignmentSet(
seq_records,
cluster_map,
consensus_threshold=args.consensus_threshold)
analysis = clustered_alignment.multiple_context_analysis(
patterns, **analysis_settings)
# write out the final clusters
clusterout_handle = file(prefix + '.clst.csv', 'w')
clustering.write(clusterout_handle)
if args.interactive:
local = copy.copy(locals())
import hyperfreq
local.update(
dict(hyperfreq=hyperfreq,
Alignment=Alignment,
AlignmentSet=AlignmentSet,
mut_pattern=mut_pattern,
write_analysis=write_analysis))
code.interact(local=local)
# Write the final analysis to file
write_analysis(analysis,
prefix,
pattern_names,
args.quants,
args.cdfs,
call_only=args.call_only)
if args.write_references:
write_reference_seqs(alignments, prefix)
# Closing files
args.alignment.close()
if args.cluster_map:
args.cluster_map.close()
def analyze(args):
import logging; logging.captureWarnings(True)
# Fetch sequence records and analysis patterns
seq_records = SeqIO.to_dict(SeqIO.parse(args.alignment, 'fasta'))
patterns = [mut_pattern.patterns[p] for p in args.patterns]
pattern_names = [p.name for p in patterns]
prefix = path.join(args.out_dir, args.prefix)
analysis_settings = dict(
rpr_cutoff=args.rpr_cutoff, significance_level=args.significance_level, quants=args.quants,
pos_quants_only=args.pos_quants_only, caller=args.caller, prior=args.prior, cdfs=args.cdfs,
quadr_maxiter=args.quadr_maxiter, optim_maxiter=args.optim_maxiter)
# Need to think about how best to fork things here; for instance, might make sense to let the user specify
# the initial clusters for whatever reason... However, specifying the reference sequences shouldn't make
# any sense there
if args.reference_sequences:
reference_sequences = SeqIO.to_dict(SeqIO.parse(args.reference_sequences, 'fasta'))
else:
reference_sequences = None
# This lets the cluster map be optional, so that this script can be used
# for naive hm filtering/analysis
cluster_map = load_cluster_map(args.cluster_map, cluster_col=args.cluster_col) if args.cluster_map else None
alignments = AlignmentSet(seq_records, cluster_map, consensus_threshold=args.consensus_threshold,
reference_sequences=reference_sequences)
# Create the analysis generator
analysis = alignments.multiple_context_analysis(patterns, **analysis_settings)
if args.cluster_threshold:
for hm_it in range(args.cluster_iterations - 1):
print " ..On hm/cluster iteration", hm_it
# Grab the HM columns from the most recent analysis and split out the pos sites
hm_columns = []
for result in analysis:
hm_columns += result['call']['mut_columns']
hm_neg_aln = Alignment(seq_records.values()).split_hypermuts(hm_columns).hm_neg_aln
# Cluster with the specified settings
clustering = alnclst.Clustering(hm_neg_aln, args.cluster_threshold,
args.consensus_threshold)
clustering = clustering.recenter(args.recentering_iterations)
clustering.merge_small_clusters(args.min_per_cluster)
cluster_map = parse_clusters(clustering.mapping_iterator(), cluster_key=0, sequence_key=1)
# Create the Alignment set
clustered_alignment = AlignmentSet(seq_records, cluster_map,
consensus_threshold=args.consensus_threshold)
analysis = clustered_alignment.multiple_context_analysis(patterns, **analysis_settings)
# write out the final clusters
clusterout_handle = file(prefix + '.clst.csv', 'w')
clustering.write(clusterout_handle)
if args.interactive:
local = copy.copy(locals())
import hyperfreq
local.update(dict(hyperfreq=hyperfreq,
Alignment=Alignment,
AlignmentSet=AlignmentSet,
mut_pattern=mut_pattern,
write_analysis=write_analysis))
code.interact(local=local)
# Write the final analysis to file
write_analysis(analysis, prefix, pattern_names, args.quants, args.cdfs, call_only=args.call_only)
if args.write_references:
write_reference_seqs(alignments, prefix)
# Closing files
args.alignment.close()
if args.cluster_map:
args.cluster_map.close()