def concatenate_alns(self):
"""Concatenate all alns into one aln.
"""
physcraper.debug("concat alns")
count = 0
for gene in self.aln_all:
if count == 0:
aln1 = self.aln_all[gene]
aln1.write(path="{}/aln1.fas".format(self.workdir),
schema="fasta")
count = 1
else:
aln2 = self.aln_all[gene]
count += 1
aln2.write(path="{}/aln{}.fas".format(self.workdir, count),
schema="fasta")
assert aln1.taxon_namespace == aln2.taxon_namespace
aln1 = DnaCharacterMatrix.concatenate([aln1, aln2])
aln1.write(path="{}/concat.fas".format(self.workdir), schema="fasta")
self.concatenated_aln = aln1
tmp_dict = {}
for taxon, seq in physcraper_obj.aln.items():
aln_dict[taxon.label] = seq
seqlen = len(seq) #should all be same bc aligned
for spp_name in spp_dict.keys():
try:
otu = random.choice(spp_dict[spp_name])
tmp_dict[spp_name] = aln_dict[otu]
except KeyError:
tmp_dict[spp_name] = "-" * seqlen
return tmp_dict
aln1 = DnaCharacterMatrix.from_dict(arbitrary_prune_fill(spp_to_otu1, gene1))
aln2 = DnaCharacterMatrix.from_dict(arbitrary_prune_fill(spp_to_otu2, gene2), taxon_namespace = aln1.taxon_namespace)
concat = DnaCharacterMatrix.concatenate([aln1,aln2])
concat.write(path="concat.fas",
schema="fasta")
#Open the two pyscraper objects
#Merge the alignements on OTT_ID?
#How to force/missing data ...
#Option 1: randomly select one seq from each ott ID.
#Option 2: Use all pairwise?
#Option 3: force mono phyly of spps? grrrrrrrrrrrrrrrr