def construct_db_of_sister(taxon, DB, outprefix):
outfile = outprefix + ".tempfa"
outfiletbl = outprefix + ".temptable"
mfid(taxon, DB, outfile, outfiletbl, remove_genomes=True)
of = open(outfiletbl, "r")
seq_taxon_dict = {}
taxon_seqs_dict = {}
for i in of:
spls = i.strip().split("\t")
try:
taxon_seqs_dict[spls[1]]
except:
taxon_seqs_dict[spls[1]] = set()
seq_taxon_dict[spls[3]] = spls[1]
taxon_seqs_dict[spls[1]].add(spls[3])
of.close()
seqs_dict = seq.read_fasta_file_return_dict(outfile)
return outfile, outfiletbl, seq_taxon_dict, taxon_seqs_dict, seqs_dict
def add_updated_seqs_to_dir(dirl, tid, DB, outfilehead, internal=False):
print(dirl)
oldids = set()
maintablename = None
for i in os.listdir(dirl):
# get the main table name so that we can update things
if i[-len(".table"):] == ".table":
maintablename = i
# read all the tables
if ".table" in i:
print("reading:", i, file=sys.stderr)
fl = open(dirl + i, "r")
for j in fl:
spls = j.strip().split("\t")
oldids.add(spls[2])
fl.close()
newids = set()
newid_info = {} #key is id, value is full bit
newid_seq = {} #key is id, value is seq
seqs, tbls = None, None
if internal == False:
seqs, tbls = mfid(tid, DB, None, None)
else:
seqs, tbls = mfid_internal(tid, DB, None, None)
for i, j in zip(seqs, tbls):
spls = j.split("\t")
newids.add(spls[2])
newid_info[spls[2]] = j
newid_seq[spls[2]] = i
diff = newids.difference(oldids)
if len(diff) > 0:
newtable = dirl + maintablename + "." + outfilehead
newseqfn = dirl + maintablename.replace(".table",
".fas") + "." + outfilehead
print("writing:", newtable, file=sys.stderr)
print("writing:", newseqfn, file=sys.stderr)
newtablef = open(newtable, "w")
newseqf = open(newseqfn, "w")
for i in diff:
newtablef.write(newid_info[i] + "\n")
newseqf.write(newid_seq[i] + "\n")
newtablef.close()
newseqf.close()
else:
print("no update", file=sys.stderr)
didntmake = set()
for i in tree.iternodes():
if "unclassified" in i.label:
didntmake.add(i)
continue
if "environmental" in i.label:
didntmake.add(i)
continue
if i.parent in didntmake:
didntmake.add(i)
continue
orig = i.label
if i != tree:
i.label = i.parent.label + "/" + i.label
tid = orig.split("_")[-1]
dirr = i.label
if len(i.children) == 0:
mfid(tid,
DB,
dirl + dirr + "/" + orig + ".fas",
dirl + dirr + "/" + orig + ".table",
True,
limitlist=taxalist)
else:
mfid_in(tid,
DB,
dirl + dirr + "/" + orig + ".fas",
dirl + dirr + "/" + orig + ".table",
True,
limitlist=taxalist)