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 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 >> fw, "\n".join(sorted(removed))
fw.close()
populate_children(opts.outfile, bestids, gffile, "gene")
def trimUTR(args):
"""
%prog trimUTR gffile
Remove UTRs in the annotation set.
If reference GFF3 is provided, reinstate UTRs from reference
transcripts after trimming.
Note: After running trimUTR, it is advised to also run
`python -m jcvi.formats.gff fixboundaries` on the resultant GFF3
to adjust the boundaries of all parent 'gene' features
"""
import gffutils
from jcvi.formats.base import SetFile
p = OptionParser(trimUTR.__doc__)
p.add_option(
"--trim5",
default=None,
type="str",
help="File containing gene list for 5' UTR trimming",
)
p.add_option(
"--trim3",
default=None,
type="str",
help="File containing gene list for 3' UTR trimming",
)
p.add_option(
"--trimrange",
default=None,
type="str",
help="File containing gene list for UTR trim back" +
"based on suggested (start, stop) coordinate range",
)
p.add_option(
"--refgff",
default=None,
type="str",
help="Reference GFF3 used as fallback to replace UTRs",
)
p.set_outfile()
opts, args = p.parse_args(args)
if len(args) != 1:
sys.exit(not p.print_help())
(gffile, ) = args
gff = make_index(gffile)
trim_both = False if (opts.trim5 or opts.trim3) else True
trim5 = SetFile(opts.trim5) if opts.trim5 else set()
trim3 = SetFile(opts.trim3) if opts.trim3 else set()
trimrange = dict()
if opts.trimrange:
trf = must_open(opts.trimrange)
for tr in trf:
assert (len(tr.split("\t")) == 3
), "Must specify (start, stop) coordinate range"
id, start, stop = tr.split("\t")
trimrange[id] = (int(start), int(stop))
trf.close()
refgff = make_index(opts.refgff) if opts.refgff else None
fw = must_open(opts.outfile, "w")
for feat in gff.iter_by_parent_childs(featuretype="gene",
order_by=("seqid", "start"),
level=1):
for c in feat:
cid, ctype, cparent = (
c.id,
c.featuretype,
c.attributes.get("Parent", [None])[0],
)
t5, t3 = False, False
if ctype == "gene":
t5 = True if cid in trim5 else False
t3 = True if cid in trim3 else False
start, end = get_cds_minmax(gff, cid)
trim(c, start, end, trim5=t5, trim3=t3, both=trim_both)
fprint(c, fw)
elif ctype == "mRNA":
utr_types, extras = [], set()
if any(id in trim5 for id in (cid, cparent)):
t5 = True
trim5.add(cid)
if any(id in trim3 for id in (cid, cparent)):
t3 = True
trim3.add(cid)
refc = None
if refgff:
try:
refc = refgff[cid]
refctype = refc.featuretype
refptype = refgff[refc.attributes["Parent"]
[0]].featuretype
if refctype == "mRNA" and refptype == "gene":
if cmp_children(cid, gff, refgff, cftype="CDS"):
reinstate(c,
refc,
trim5=t5,
trim3=t3,
both=trim_both)
if t5:
utr_types.append("five_prime_UTR")
if t3:
utr_types.append("three_prime_UTR")
for utr_type in utr_types:
for utr in refgff.children(
refc, featuretype=utr_type):
extras.add(utr)
for exon in refgff.region(
region=utr,
featuretype="exon"):
if exon.attributes["Parent"][
0] == cid:
extras.add(exon)
else:
refc = None
except gffutils.exceptions.FeatureNotFoundError:
pass
start, end = get_cds_minmax(gff, cid, level=1)
if cid in trimrange:
start, end = range_minmax([trimrange[cid], (start, end)])
if not refc:
trim(c, start, end, trim5=t5, trim3=t3, both=trim_both)
fprint(c, fw)
for cc in gff.children(cid, order_by="start"):
_ctype = cc.featuretype
if _ctype not in utr_types:
if _ctype != "CDS":
if _ctype == "exon":
eskip = [
range_overlap(to_range(cc), to_range(x))
for x in extras if x.featuretype == "exon"
]
if any(eskip):
continue
trim(cc,
start,
end,
trim5=t5,
trim3=t3,
both=trim_both)
fprint(cc, fw)
else:
fprint(cc, fw)
for x in extras:
fprint(x, fw)
fw.close()
def trimUTR(args):
"""
%prog trimUTR gffile
Remove UTRs in the annotation set.
If reference GFF3 is provided, reinstate UTRs from reference
transcripts after trimming.
Note: After running trimUTR, it is advised to also run
`python -m jcvi.formats.gff fixboundaries` on the resultant GFF3
to adjust the boundaries of all parent 'gene' features
"""
import gffutils
from jcvi.formats.base import SetFile
p = OptionParser(trimUTR.__doc__)
p.add_option("--trim5", default=None, type="str", \
help="File containing gene list for 5' UTR trimming")
p.add_option("--trim3", default=None, type="str", \
help="File containing gene list for 3' UTR trimming")
p.add_option("--trimrange", default=None, type="str", \
help="File containing gene list for UTR trim back" + \
"based on suggested (start, stop) coordinate range")
p.add_option("--refgff", default=None, type="str", \
help="Reference GFF3 used as fallback to replace UTRs")
p.set_outfile()
opts, args = p.parse_args(args)
if len(args) != 1:
sys.exit(not p.print_help())
gffile, = args
gff = make_index(gffile)
trim_both = False if (opts.trim5 or opts.trim3) else True
trim5 = SetFile(opts.trim5) if opts.trim5 else set()
trim3 = SetFile(opts.trim3) if opts.trim3 else set()
trimrange = dict()
if opts.trimrange:
trf = must_open(opts.trimrange)
for tr in trf:
assert len(tr.split("\t")) == 3, \
"Must specify (start, stop) coordinate range"
id, start, stop = tr.split("\t")
trimrange[id] = (int(start), int(stop))
trf.close()
refgff = make_index(opts.refgff) if opts.refgff else None
fw = must_open(opts.outfile, "w")
for feat in gff.iter_by_parent_childs(featuretype="gene", order_by=("seqid", "start"), level=1):
for c in feat:
cid, ctype, cparent = c.id, c.featuretype, \
c.attributes.get('Parent', [None])[0]
t5, t3 = False, False
if ctype == "gene":
t5 = True if cid in trim5 else False
t3 = True if cid in trim3 else False
start, end = get_cds_minmax(gff, cid)
trim(c, start, end, trim5=t5, trim3=t3, both=trim_both)
fprint(c, fw)
elif ctype == "mRNA":
utr_types, extras = [], set()
if any(id in trim5 for id in (cid, cparent)):
t5 = True
trim5.add(cid)
if any(id in trim3 for id in (cid, cparent)):
t3 = True
trim3.add(cid)
refc = None
if refgff:
try:
refc = refgff[cid]
refctype = refc.featuretype
refptype = refgff[refc.attributes['Parent'][0]].featuretype
if refctype == "mRNA" and refptype == "gene":
if cmp_children(cid, gff, refgff, cftype="CDS"):
reinstate(c, refc, trim5=t5, trim3=t3, both=trim_both)
if t5: utr_types.append('five_prime_UTR')
if t3: utr_types.append('three_prime_UTR')
for utr_type in utr_types:
for utr in refgff.children(refc, featuretype=utr_type):
extras.add(utr)
for exon in refgff.region(region=utr, featuretype="exon"):
if exon.attributes['Parent'][0] == cid:
extras.add(exon)
else:
refc = None
except gffutils.exceptions.FeatureNotFoundError:
pass
start, end = get_cds_minmax(gff, cid, level=1)
if cid in trimrange:
start, end = range_minmax([trimrange[cid], (start, end)])
if not refc:
trim(c, start, end, trim5=t5, trim3=t3, both=trim_both)
fprint(c, fw)
for cc in gff.children(cid, order_by=("start")):
_ctype = cc.featuretype
if _ctype not in utr_types:
if _ctype != "CDS":
if _ctype == "exon":
eskip = [range_overlap(to_range(cc), to_range(x)) \
for x in extras if x.featuretype == 'exon']
if any(skip for skip in eskip): continue
trim(cc, start, end, trim5=t5, trim3=t3, both=trim_both)
fprint(cc, fw)
else:
fprint(cc, fw)
for x in extras:
fprint(x, fw)
fw.close()