def augustus(args):
"""
%prog augustus augustus.gff3 > reformatted.gff3
AUGUSTUS does generate a gff3 (--gff3=on) but need some refinement.
"""
from jcvi.formats.gff import Gff
p = OptionParser(augustus.__doc__)
p.set_outfile()
opts, args = p.parse_args(args)
if len(args) != 1:
sys.exit(not p.print_help())
ingff3, = args
gff = Gff(ingff3)
fw = must_open(opts.outfile, "w")
seen = defaultdict(int)
for g in gff:
if g.type not in ("gene", "transcript", "CDS"):
continue
if g.type == "transcript":
g.type = "mRNA"
prefix = g.seqid + "_"
pid = prefix + g.id
newid = "{0}-{1}".format(pid, seen[pid]) if pid in seen else pid
seen[pid] += 1
g.attributes["ID"] = [newid]
g.attributes["Parent"] = [(prefix + x) for x in g.attributes["Parent"]]
g.update_attributes()
print >> fw, g
fw.close()
def trimUTR(args):
"""
%prog trimUTR gffile
Remove UTRs in the annotation set.
"""
p = OptionParser(trimUTR.__doc__)
p.set_outfile()
opts, args = p.parse_args(args)
if len(args) != 1:
sys.exit(not p.print_help())
gffile, = args
g = make_index(gffile)
gff = Gff(gffile)
mRNA_register = {}
fw = must_open(opts.outfile, "w")
for c in gff:
cid, ctype = c.accn, c.type
if ctype == "gene":
start, end = get_cds_minmax(g, cid)
trim(c, start, end)
elif ctype == "mRNA":
start, end = get_cds_minmax(g, cid, level=1)
trim(c, start, end)
mRNA_register[cid] = (start, end)
elif ctype != "CDS":
start, end = mRNA_register[c.parent]
trim(c, start, end)
if c.start > c.end:
print >> sys.stderr, cid, \
"destroyed [{0} > {1}]".format(c.start, c.end)
else:
print >> fw, c
def pasa(args):
"""
%prog ${pasadb}.assemblies.fasta ${pasadb}.pasa_assemblies.gff3
Wraps `pasa_asmbls_to_training_set.dbi`.
"""
from jcvi.formats.base import SetFile
from jcvi.formats.gff import Gff
p = OptionParser(pasa.__doc__)
p.set_home("pasa")
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
fastafile, gffile = args
transcodergff = fastafile + ".transdecoder.gff3"
transcodergenomegff = fastafile + ".transdecoder.genome.gff3"
if need_update((fastafile, gffile), (transcodergff, transcodergenomegff)):
cmd = "{0}/scripts/pasa_asmbls_to_training_set.dbi".format(
opts.pasa_home)
cmd += " --pasa_transcripts_fasta {0} --pasa_transcripts_gff3 {1}".\
format(fastafile, gffile)
sh(cmd)
completeids = fastafile.rsplit(".", 1)[0] + ".complete.ids"
if need_update(transcodergff, completeids):
cmd = "grep complete {0} | cut -f1 | sort -u".format(transcodergff)
sh(cmd, outfile=completeids)
complete = SetFile(completeids)
seen = set()
completegff = transcodergenomegff.rsplit(".", 1)[0] + ".complete.gff3"
fw = open(completegff, "w")
gff = Gff(transcodergenomegff)
for g in gff:
a = g.attributes
if "Parent" in a:
id = a["Parent"][0]
else:
id = a["ID"][0]
asmbl_id = id.split("|")[0]
if asmbl_id not in complete:
continue
print >> fw, g
if g.type == "gene":
seen.add(id)
fw.close()
logging.debug("A total of {0} complete models extracted to `{1}`.".\
format(len(seen), completegff))
def yeasttruth(args):
"""
%prog yeasttruth Pillars.tab *.gff
Prepare pairs data for 14 yeasts.
"""
p = OptionParser(yeasttruth.__doc__)
p.set_outfile()
opts, args = p.parse_args(args)
if len(args) < 2:
sys.exit(not p.print_help())
pillars = args[0]
gffiles = args[1:]
aliases = {}
pivot = {}
for gffile in gffiles:
is_pivot = op.basename(gffile).startswith("Saccharomyces_cerevisiae")
gff = Gff(gffile)
for g in gff:
if g.type != "gene":
continue
for a in g.attributes["Alias"]:
aliases[a] = g.accn
if is_pivot:
pivot[a] = g.accn
logging.debug("Aliases imported: {0}".format(len(aliases)))
logging.debug("Pivot imported: {0}".format(len(pivot)))
fw = open("yeast.aliases", "w")
for k, v in sorted(aliases.items()):
print("\t".join((k, v)), file=fw)
fw.close()
fp = open(pillars)
pairs = set()
fw = must_open(opts.outfile, "w")
for row in fp:
atoms = [x for x in row.split() if x != "---"]
pps = [pivot[x] for x in atoms if x in pivot]
atoms = [aliases[x] for x in atoms if x in aliases]
for p in pps:
for a in atoms:
if p == a:
continue
pairs.add(tuple(sorted((p, a))))
for a, b in sorted(pairs):
print("\t".join((a, b)), file=fw)
fw.close()
def uniq(args):
"""
%prog uniq gffile cdsfasta
Remove overlapping gene models. Similar to formats.gff.uniq(), overlapping
'piles' are processed, one by one.
Here, we use a different algorithm, that retains the best non-overlapping
subset witin each pile, rather than single best model. Scoring function is
also different, rather than based on score or span, we optimize for the
subset that show the best combined score. Score is defined by:
score = (1 - AED) * length
"""
p = OptionParser(uniq.__doc__)
p.set_outfile()
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
gffile, cdsfasta = args
gff = Gff(gffile)
sizes = Sizes(cdsfasta).mapping
gene_register = {}
for g in gff:
if g.type != "mRNA":
continue
aed = float(g.attributes["_AED"][0])
gene_register[g.parent] = (1 - aed) * sizes[g.accn]
allgenes = import_feats(gffile)
g = get_piles(allgenes)
bestids = set()
for group in g:
ranges = [
to_range(x, score=gene_register[x.accn], id=x.accn) for x in group
]
selected_chain, score = range_chain(ranges)
bestids |= set(x.id for x in selected_chain)
removed = set(x.accn for x in allgenes) - bestids
fw = open("removed.ids", "w")
print("\n".join(sorted(removed)), file=fw)
fw.close()
populate_children(opts.outfile, bestids, gffile, "gene")
def get_cds_beds(gffile, noUTR=False):
from jcvi.formats.gff import Gff
mrnabed = None
cdsbeds = []
gf = Gff(gffile)
for g in gf:
if g.type == "mRNA":
mrnabed = g.bedline
elif g.type == "CDS":
cdsbeds.append(g.bedline)
if noUTR:
mrnabed.start = min(x.start for x in cdsbeds)
mrnabed.end = max(x.end for x in cdsbeds)
return mrnabed, cdsbeds
def augustus(args):
"""
%prog augustus augustus.gff3 > reformatted.gff3
AUGUSTUS does generate a gff3 (--gff3=on) but need some refinement.
"""
p = OptionParser(augustus.__doc__)
opts, args = p.parse_args(args)
if len(args) != 1:
sys.exit(not p.print_help())
ingff3, = args
gff = Gff(ingff3)
for g in gff:
if g.type not in ("gene", "transcript", "CDS"):
continue
if g.type == "transcript":
g.type = "mRNA"
print g