def get_concatenated_alg(alg_filenames, models=None,
sp_field=0, sp_delimiter="_",
kill_thr=0.0,
keep_species=set()):
# Concat alg container
concat = SeqGroup()
# Used to store different model partitions
concat.id2partition = {}
if not models:
models = ["None"]*len(alg_filenames)
else:
if len(models) != len(alg_filenames):
raise ValueError("Different number of algs and model names was found!")
expected_total_length = 0
# Check algs and gets the whole set of species
alg_objects = []
sp2alg = defaultdict(list)
for algfile, matrix in zip(alg_filenames, models):
alg = SeqGroup(algfile, "iphylip_relaxed")
alg_objects.append(alg)
lenseq = None
browsed_species = set()
alg.sp2seq = {}
# Set best matrix for this alignment
alg.matrix = matrix
# Change seq names to contain only species names
for i, seq in alg.id2seq.iteritems():
name = db.get_seq_name(alg.id2name[i])
taxid = get_species_code(name, splitter=sp_delimiter, field=sp_field)
if lenseq is not None and len(seq) != lenseq:
raise Exception("Inconsistent alignment when concatenating: Unequal length")
elif lenseq is None:
lenseq = len(seq)
alg.seqlength = len(seq)
expected_total_length += len(seq)
if taxid in browsed_species:
raise Exception("Inconsistent alignment when concatenating: Repeated species")
browsed_species.add(taxid) # Check no duplicated species in the same alg
sp2alg[taxid].append(alg) # Records all species seen in all algs.
alg.sp2seq[taxid] = seq
valid_species = [sp for sp in sp2alg.iterkeys() \
if sp in keep_species or \
len(sp2alg[sp])/float(len(alg_objects)) > kill_thr]
log.info("%d out of %d will be kept (missing factor threshold=%g, %d species forced to kept)" %\
(len(valid_species), len(sp2alg), kill_thr, len(keep_species)))
def sort_single_algs(alg1, alg2):
r = cmp(alg1.matrix, alg2.matrix)
if r == 0:
return cmp(sorted(alg1.id2name.values()),
sorted(alg2.id2name.values()))
else:
return r
sorted_algs = sorted(alg_objects, sort_single_algs)
concat_alg_lengths = [alg.seqlength for alg in sorted_algs]
model2win = {}
model2size = {}
for alg in sorted_algs:
model2size[alg.matrix] = model2size.get(alg.matrix, 0) + alg.seqlength
# Create concat alg
concat.id2seq = defaultdict(list)
for sp in sorted(valid_species):
log.log(20, "Concatenating sequences of [%s]" %sp)
for alg in sorted_algs:
seq = alg.sp2seq.get(sp, "-" * alg.seqlength)
concat.id2seq[sp].append(seq)
#current_seq = concat.id2seq.get(sp, "")
#concat.id2seq[sp] = current_seq + seq.strip()
concat.id2name[sp] = sp
concat.name2id[sp] = sp
concat.id2comment[sp] = [""]
concat.id2seq[sp] = ''.join(concat.id2seq[sp])
current_pos = 0
partitions = []
for model in sorted(model2size.keys()):
size = model2size[model]
part = "%s, %s = %d-%d" % (model, model+"_genes", \
current_pos + 1,\
current_pos + size)
current_pos += size
partitions.append(part)
# Basic Checks
seq_sizes = [len(seq) for seq in concat.id2seq.values()]
if len(set(seq_sizes)) != 1:
raise Exception("Concatenated alignment is not consistent: unequal seq length ")
if seq_sizes[0] != expected_total_length:
raise Exception("The size of concatenated alg is not what expected")
return concat, partitions, sp2alg, valid_species, concat_alg_lengths