def fold_exterior_loop(seq, con, exterior_only=True):
""" Constrained folding
The default behavior is "exterior_only", which replaces all constrained
helices with short 'NNN' stretches at the sequence level. This reduces
the sequence length (n) and therefore the runtime O(n^3)
:param seq: RNA sequence
:param con: constraint
:param exterior_only: only fold the extior loop region
:return: secondary structure
"""
if exterior_only:
spacer = 'NNN'
pt = ril.make_pair_table(con, base=0)
ext = ''
# shrink the sequcnes
skip = 0
for i, j in enumerate(pt):
if i < skip: continue
if j == -1:
ext += seq[i]
else:
ext += spacer
skip = j + 1
css, cfe = RNA.fold(ext)
# replace characters in constraint
c, skip = 0, 0
for i, j in enumerate(pt):
if i < skip: continue
if j == -1:
con = con[:i] + css[c] + con[i + 1:]
c += 1
else:
c += len(spacer)
skip = j + 1
ss = con
else:
# Force copy of string for ViennaRNA swig interface bug
tmp = (con + '.')[:-1]
RNA.cvar.fold_constrained = 1
ss, mfe = RNA.fold(seq, tmp)
RNA.cvar.fold_constrained = 0
return ss
def open_breathing_helices(seq, ss, free=6):
""" open all breathable helices, i.e. those that share a base-pair
with an exterior loop region
"""
nbrs = set()
pt = ril.make_pair_table(ss, base=0)
# mutable secondary structure
nbr = list(ss)
rec_fill_nbrs(nbrs, ss, nbr, pt, (0, len(ss)), free)
nbrs.add(''.join(nbr))
return nbrs
def open_breathing_helices(seq, ss, free=6):
""" open all breathable helices, i.e. those that share a base-pair
with an exterior loop region
"""
nbrs = set()
pt = ril.make_pair_table(ss, base=0)
# mutable secondary structure
nbr = list(ss)
rec_fill_nbrs(nbrs, ss, nbr, pt, (0, len(ss)), free)
nbrs.add(''.join(nbr))
return nbrs
def aptamer_energy(
seq,
ss,
verb=False,
# Default Theophylline
apt='GAUACCAG' + '&' + 'CCCUUGGCAGC',
poc='(...((((' + '&' + ')...)))...)',
bfe=-8.86): # at 25*C; -9.22 at 37*C
"""
Check if a sequence/structure pair contains the
ligand binding pocket (apt/poc). If so, return the
binding free energy (bfe), otherwise return 0.
Multiple pockets will return bfe multiple times!
TODO: allow hairpin pockets (e.g. tetracycline)
"""
[aptL, aptR] = apt.split('&')
[pocL, pocR] = poc.split('&')
patL = re.compile(aptL)
patR = re.compile(aptR)
sites = 0
for mL in patL.finditer(seq):
if pocL != ss[mL.start():mL.end()]:
continue
for mR in patR.finditer(seq):
if mR.start() < mL.end():
continue
if pocR != ss[mR.start():mR.end()]:
continue
# Now make sure that these are really base-pairs
ptable = ril.make_pair_table(ss, base=0)
if mL.start() == ptable[mR.end() - 1] and \
mL.end() - 1 == ptable[mR.start()] and \
mR.start() == ptable[mL.end() - 1] and \
mR.end() - 1 == ptable[mL.start()]:
# if verb :
# print >> sys.stderr, "{:s} {:6.2f}".format(ss, bfe)
sites += 1
return bfe * sites
def aptamer_energy(seq, ss, verb=False,
# Default Theophylline
apt='GAUACCAG' + '&' + 'CCCUUGGCAGC',
poc='(...((((' + '&' + ')...)))...)',
bfe=-8.86): # at 25*C; -9.22 at 37*C
"""
Check if a sequence/structure pair contains the
ligand binding pocket (apt/poc). If so, return the
binding free energy (bfe), otherwise return 0.
Multiple pockets will return bfe multiple times!
TODO: allow hairpin pockets (e.g. tetracycline)
"""
[aptL, aptR] = apt.split('&')
[pocL, pocR] = poc.split('&')
patL = re.compile(aptL)
patR = re.compile(aptR)
sites = 0
for mL in patL.finditer(seq):
if pocL != ss[mL.start():mL.end()]:
continue
for mR in patR.finditer(seq):
if mR.start() < mL.end():
continue
if pocR != ss[mR.start():mR.end()]:
continue
# Now make sure that these are really base-pairs
ptable = ril.make_pair_table(ss, base=0)
if mL.start() == ptable[mR.end() - 1] and \
mL.end() - 1 == ptable[mR.start()] and \
mR.start() == ptable[mL.end() - 1] and \
mR.end() - 1 == ptable[mL.start()]:
# if verb :
# print >> sys.stderr, "{:s} {:6.2f}".format(ss, bfe)
sites += 1
return bfe * sites
def fold_exterior_loop(md, seq, con, ext_moves):
""" Constrained folding of the exterior loop.
All constrained helices are replaced with the motif:
NNNNNNN
((xxx))
for example a helix with the closing-stack CG-UG:
CG ~ UG -> CGNNNUG
(( ~ )) -> ((xxx))
This reduces the sequence length (n) and therefore the runtime O(n^3),
and it enables the identification of independent structures with the same
exterior loop features.
Args:
md (RNA.md()): ViennaRNA model details (temperature, noLP, etc.)
seq (str): RNA sequence
con (str): RNA structure constraint
ext_moves (dict()): Dictionary storing all mappings from exterior-loop
constraints (features) to parents.
Returns:
(str, str):
"""
spacer = 'NNN'
pt = ril.make_pair_table(con, base=0)
ext_seq = ''
ext_con = ''
# shrink the sequences
skip = 0
for i, j in enumerate(pt):
if i < skip:
continue
if j == -1:
ext_seq += seq[i]
ext_con += '.'
else:
ext_seq += seq[i] + seq[i + 1]
ext_seq += spacer
ext_seq += seq[j - 1] + seq[j]
ext_con += '(('
ext_con += 'x' * len(spacer)
ext_con += '))'
skip = j + 1
# If we have seen this exterior loop before, then we don't need to
# calculate again, and we have to trace back if the parents are connected.
if ext_seq in ext_moves:
css = ext_moves[ext_seq][1]
else:
fc_tmp = RNA.fold_compound(ext_seq, md)
fc_tmp.constraints_add(
ext_con, RNA.CONSTRAINT_DB_DEFAULT | RNA.CONSTRAINT_DB_ENFORCE_BP)
css, cfe = fc_tmp.mfe()
ext_moves[ext_seq] = [set(), css]
del fc_tmp
# replace characters in constraint
c, skip = 0, 0
for i, j in enumerate(pt):
if i < skip:
continue
if j == -1:
con = con[:i] + css[c] + con[i + 1:]
c += 1
else:
c += len(spacer) + 4
skip = j + 1
ss = con
return ss, ext_seq
def fold_exterior_loop(md, seq, con, ext_moves):
""" Constrained folding of the exterior loop.
All constrained helices are replaced with the motif:
NNNNNNN
((xxx))
for example a helix with the closing-stack CG-UG:
CG ~ UG -> CGNNNUG
(( ~ )) -> ((xxx))
This reduces the sequence length (n) and therefore the runtime O(n^3),
and it enables the identification of independent structures with the same
exterior loop features.
Args:
md (RNA.md()): ViennaRNA model details (temperature, noLP, etc.)
seq (str): RNA sequence
con (str): RNA structure constraint
ext_moves (dict()): Dictionary storing all mappings from exterior-loop
constraints (features) to parents.
Returns:
(str, str):
"""
spacer = 'NNN'
pt = ril.make_pair_table(con, base=0)
ext_seq = ''
ext_con = ''
# shrink the sequences
skip = 0
for i, j in enumerate(pt):
if i < skip:
continue
if j == -1:
ext_seq += seq[i]
ext_con += '.'
else:
ext_seq += seq[i] + seq[i + 1]
ext_seq += spacer
ext_seq += seq[j - 1] + seq[j]
ext_con += '(('
ext_con += 'x' * len(spacer)
ext_con += '))'
skip = j + 1
# If we have seen this exterior loop before, then we don't need to
# calculate again, and we have to trace back if the parents are connected.
if ext_seq in ext_moves:
css = ext_moves[ext_seq][1]
else:
fc_tmp = RNA.fold_compound(ext_seq, md)
fc_tmp.constraints_add(
ext_con, RNA.CONSTRAINT_DB_DEFAULT | RNA.CONSTRAINT_DB_ENFORCE_BP)
css, cfe = fc_tmp.mfe()
ext_moves[ext_seq] = [set(), css]
del fc_tmp
# replace characters in constraint
c, skip = 0, 0
for i, j in enumerate(pt):
if i < skip:
continue
if j == -1:
con = con[:i] + css[c] + con[i + 1:]
c += 1
else:
c += len(spacer) + 4
skip = j + 1
ss = con
return ss, ext_seq