def save(session, alignment, stream):
print("CLUSTAL W ALN saved from UCSF ChimeraX", file=stream)
print("", file=stream)
max_name = max([len(seq.name) for seq in alignment.seqs])
name_format = "%%-%ds" % (max_name+5)
from chimerax.atomic import Sequence
aln_len = len(alignment.seqs[0])
for start in range(0, aln_len, LINELEN):
end = min(aln_len, start + LINELEN)
for seq in alignment.seqs:
name = seq.name.replace(' ', '_')
temp_seq = Sequence()
temp_seq.extend(seq[start:end])
if len(temp_seq.ungapped()) == 0:
print(name_format % name, seq[start:end], file=stream)
else:
temp_seq = Sequence()
temp_seq.extend(seq[:end])
print(name_format % name, seq[start:end], len(temp_seq.ungapped()), file=stream)
from .. import clustal_strong_groups, clustal_weak_groups
conservation = []
for pos in range(start, end):
# completely conserved?
first = alignment.seqs[0][pos].upper()
if first.isupper():
for seq in alignment.seqs[1:]:
if seq[pos].upper() != first:
break
else:
# conserved
conservation.append('*')
continue
# "strongly"/"weakly" conserved?
conserved = False
for groups, character in [(clustal_strong_groups, ':'), (clustal_weak_groups, '.')]:
for group in groups:
for seq in alignment.seqs:
if seq[pos].upper() not in group:
break
else:
# conserved
conserved = True
break
if conserved:
conservation.append(character)
break
if not conserved:
# remainder
conservation.append(' ')
print(name_format % " ", "".join(conservation), file=stream)
print("", file=stream)
def regularized_seq(aseq, chain):
mmap = aseq.match_maps[chain]
from .common import modeller_copy
rseq = modeller_copy(aseq)
rseq.description = "structure:" + chain_save_name(chain)
seq_chars = list(rseq.characters)
from chimerax.atomic import Sequence
from chimerax.pdb import standard_polymeric_res_names as std_res_names
in_seq_hets = []
num_res = 0
for ungapped in range(len(aseq.ungapped())):
gapped = aseq.ungapped_to_gapped(ungapped)
if ungapped not in mmap:
seq_chars[gapped] = '-'
else:
r = mmap[ungapped]
num_res += 1
if r.name not in std_res_names:
in_seq_hets.append(r.name)
seq_chars[gapped] = '.'
else:
seq_chars[gapped] = Sequence.rname3to1(mmap[ungapped].name)
s = chain.structure
het_set = getattr(s, 'in_seq_hets', set())
# may want to preserve all-HET chains, so don't auto-exclude them
if num_res != len(in_seq_hets):
het_set.update(in_seq_hets)
s.in_seq_hets = het_set
rseq.characters = "".join(seq_chars)
return rseq
def show_mav(self, ids):
# Collect names and sequences of selected matches.
# All sequences should have the same length because
# they include gaps generated from BLAST alignment.
ids.insert(0, 0)
names = []
seqs = []
for sid in ids:
name, seq = self._sequences[sid]
names.append(name)
seqs.append(seq)
# Find columns that are gaps in all sequences and remove them.
all_gaps = set()
for i in range(len(seqs[0])):
for seq in seqs:
if seq[i].isalpha():
break
else:
all_gaps.add(i)
if all_gaps:
for i in range(len(seqs)):
seq = seqs[i]
new_seq = ''.join(
[seq[n] for n in range(len(seq)) if n not in all_gaps])
seqs[i] = new_seq
# Generate multiple sequence alignment file
# Ask sequence viewer to display alignment
from chimerax.atomic import Sequence
seqs = [
Sequence(name=name, characters=seqs[i])
for i, name in enumerate(names)
]
name = "%s [%d]" % (self._instance_name, self._viewer_index)
self.session.alignments.new_alignment(seqs, name)
def read(session, f):
# skip header crap
in_header = True
line_num = 0
sequences = []
for line in f.readlines():
line = line.strip()
line_num += 1
if not line:
continue
fields = line.split()
if in_header:
if len(fields[0]) == 2:
continue
if fields[0].startswith('#='):
# some Pfam seed alignments have undocumented #=RF header
continue
in_header = False
if len(fields) != 2:
raise FormatSyntaxError(
"Sequence line %d not of form 'seq-name seq-letters'" % line_num)
seq = Sequence(name=make_readable(fields[0]))
seq.extend(fields[1])
sequences.append(seq)
f.close()
return sequences, {}, {}
def read(session, f):
want = 'init'
sequences = []
for line in f.readlines():
line = line.strip()
if want == 'init':
if len(line) < 4:
continue
if line[0] != '>' or line[3] != ';':
continue
sequences.append(Sequence(name=make_readable(line[4:])))
pir_type = line[1:3]
if pir_type in ("P1", "F1"):
sequences[-1].nucleic = True
else:
sequences[-1].nucleic = False
sequences[-1].pir_type = pir_type
want = 'description'
elif want == 'description':
sequences[-1].description = line
sequences[-1].pir_description = line
want = 'sequence'
elif want == 'sequence':
if not line:
continue
if line[-1] == '*':
want = 'init'
line = line[:-1]
sequences[-1].extend("".join([c for c in line if not c.isspace()]))
f.close()
if want != 'init':
raise FormatSyntaxError("Could not find end of sequence '%s'" %
sequences[-1].name)
return sequences, {}, {}
def find_affixes(chains, chain_info):
from chimerax.pdb import standard_polymeric_res_names as std_res_names
in_seq_hets = []
prefixes = []
suffixes = []
from chimerax.atomic import Sequence
for chain in chains:
try:
aseq, target = chain_info[chain]
except KeyError:
prefixes.append('')
suffixes.append('')
continue
match_map = aseq.match_maps[chain]
prefix = ''
for r in chain.existing_residues:
if r in match_map:
break
if r.name not in std_res_names:
in_seq_hets.append(r.name)
prefix += '.'
else:
prefix += Sequence.rname3to1(r.name)
prefixes.append(prefix)
suffix = ''
for r in reversed(chain.existing_residues):
if r in match_map:
break
if r.name not in std_res_names:
in_seq_hets.append(r.name)
suffix = '.' + suffix
else:
suffix = Sequence.rname3to1(r.name) + suffix
suffixes.append(suffix)
s = chain.structure
het_set = getattr(s, 'in_seq_hets', set())
het_set.update(in_seq_hets)
s.in_seq_hets = het_set
return prefixes, suffixes
def seqalign_chain(session, chains):
'''
Show chain sequence(s)
Parameters
----------
chains : list of Chain
Chains to show
'''
if len(chains) == 1:
chain = chains[0]
ident = ".".join([str(part) for part in chain.structure.id]) + "/" + chain.chain_id
alignment = session.alignments.new_alignment([chain], ident, seq_viewer="sv",
auto_associate=None, intrinsic=True)
else:
# all chains have to have the same sequence, and they will all be associated with
# that sequence
sequences = set([chain.characters for chain in chains])
if len(sequences) != 1:
raise UserError("Chains must have same sequence")
chars = sequences.pop()
chain_ids = set([chain.chain_id for chain in chains])
if len(chain_ids) < len(chains) or len(chain_ids) > 10:
name = "%d chains" % len(chains)
else:
name = "chains %s" % ",".join(sorted(list(chain_ids)))
from chimerax.atomic import Sequence
seq = Sequence(name=name, characters=chars)
def get_numbering_start(chain):
for i, r in enumerate(chain.residues):
if r is None or r.deleted:
continue
return r.number - i
return None
starts = set([get_numbering_start(chain) for chain in chains])
starts.discard(None)
if len(starts) == 1:
seq.numbering_start = starts.pop()
alignment = session.alignments.new_alignment([seq], None, seq_viewer="sv",
auto_associate=False, name=chains[0].description, intrinsic=True)
alignment.suspend_notify_observers()
for chain in chains:
alignment.associate(chain, keep_intrinsic=True)
alignment.resume_notify_observers()
def nw_assoc(session, align_seq, struct_seq):
'''Wrapper around Needleman-Wunsch matching, to make it return the same kinds of values
that try_assoc returns'''
from chimerax.atomic import Sequence, SeqMatchMap
sseq = struct_seq
aseq = Sequence(name=align_seq.name, characters=align_seq.ungapped())
aseq.circular = align_seq.circular
from chimerax.alignment_algs.NeedlemanWunsch import nw
score, match_list = nw(sseq, aseq)
errors = 0
# matched are in reverse order...
try:
m_end = match_list[0][0]
except IndexError:
m_end = -1
if m_end < len(sseq) - 1:
# trailing unmatched
errors += len(sseq) - m_end - 1
match_map = SeqMatchMap(align_seq, struct_seq)
last_match = m_end + 1
for s_index, a_index in match_list:
if sseq[s_index] != aseq[a_index]:
errors += 1
if s_index < last_match - 1:
# gap in structure sequence
errors += last_match - s_index - 1
res = sseq.residues[s_index]
if res:
match_map.match(res, a_index)
last_match = s_index
if last_match > 0:
# beginning unmatched
errors += last_match
if len(sseq) > len(aseq):
# unmatched residues forced, reduce errors by that amount...
errors -= len(sseq) - len(aseq)
return match_map, errors
def fetch_uniprot(session, ident, ignore_cache=False):
'Fetch UniProt data'
from chimerax.core.errors import UserError, CancelOperation
try:
accession = map_uniprot_ident(ident)
seq_string, full_name, features = fetch_uniprot_accession_info(session, accession,
ignore_cache=ignore_cache)
except InvalidAccessionError as e:
raise UserError(str(e))
except CancelOperation:
session.logger.status("Fetch of %s cancelled" % ident)
return
from chimerax.atomic import Sequence
seq = Sequence(name=ident)
seq.extend(seq_string)
session.logger.status("Opening UniProt %s" % ident)
session.alignments.new_alignment([seq], ident)
return [], "Opened UniProt %s" % ident
def read(session, f):
in_header = True
sequences = []
line_num = 0
for line in f.readlines():
line_num += 1
if in_header:
if line.startswith("CLUSTAL"):
in_header = False
first_block = True
else:
if line.strip() != "":
raise FormatSyntaxError(
"First non-blank line does not start with 'CLUSTAL'")
continue
if not line or line[0].isspace():
if sequences:
first_block = False
expect = 0
continue
try:
seq_name, seq_block, num_residues = line.split()
except ValueError:
try:
seq_name, seq_block = line.strip().split()
except ValueError:
raise FormatSyntaxError(
"Line %d is not sequence name followed by sequence "
"contents and optional ungapped length" % line_num)
if first_block:
sequences.append(Sequence(name=make_readable(seq_name)))
sequences[-1].append(seq_block)
continue
try:
seq = sequences[expect]
except IndexError:
raise FormatSyntaxError(
"Sequence on line %d not in initial sequence block" % line_num)
expect += 1
seq.append(seq_block)
f.close()
return sequences, {}, {}
def _read_sequences(self, f):
from chimerax.atomic import Sequence
self.sequence_list = []
while 1:
line = f.readline()
if not line:
raise FormatSyntaxError('no alignment separator')
if line == '//\n' or line == '//\r\n':
break
m = MSF._Sum.match(line)
if m is not None:
name = m.group(1)
length = m.group(2)
check = m.group(3)
weight = m.group(4)
s = Sequence(name=make_readable(name))
self.sequence_list.append(s)
s.attrs = {}
s.attrs['MSF length'] = length
s.attrs['MSF check'] = check
s.attrs['MSF weight'] = weight
if not self.sequence_list:
raise FormatSyntaxError('No sequences found in header')
def read(session, f):
from chimerax.atomic import Sequence
from ..parse import FormatSyntaxError, make_readable
in_sequence = False
sequences = []
for line in f.readlines():
if in_sequence:
if not line or line.isspace():
in_sequence = False
continue
if line[0] == '>':
in_sequence = False
# fall through
else:
sequences[-1].extend(line.strip())
if not in_sequence:
if line[0] == '>':
if sequences and len(sequences[-1]) == 0:
raise FormatSyntaxError("No sequence found for %s"
% sequences[-1].name)
in_sequence = True
sequences.append(Sequence(name=make_readable(line[1:])))
return sequences, {}, {}
def read(session, f):
line_num = 0
file_attrs = {}
file_markups = {}
seq_attrs = {}
seq_markups = {}
sequences = {}
seq_sequence = []
for line in f.readlines():
line = line.rstrip() # drop trailing newline/whitespace
line_num += 1
if line_num == 1:
if line.startswith("# STOCKHOLM"):
continue
raise FormatSymtaxError("File does not start with '# STOCKHOLM'")
if not line:
continue
if line.startswith('#='):
markup_type = line[2:4]
markup = line[5:].strip()
def try_split(num_split):
fields = markup.split(None, num_split)
if len(fields) == num_split:
# value is empty
fields.append("")
if len(fields) != num_split + 1:
raise FormatSyntaxError(
"Not enough arguments after #=%s markup on line %d" %
(markup_type, line_num))
return fields
if markup_type == "GF":
tag, val = try_split(1)
tag = tag.replace("_", " ")
tag = generic_file_attrs.get(tag, "Stockholm " + tag)
if tag in file_attrs:
file_attrs[tag] += '\n' + val
else:
file_attrs[tag] = val
elif markup_type == "GS":
seq_name, tag, val = try_split(2)
tag = tag.replace("_", " ")
attrs = seq_attrs.setdefault(seq_name, {})
tag = generic_seq_attrs.get(tag, "Stockholm " + tag)
if tag in attrs:
attrs[tag] += '\n' + val
else:
attrs[tag] = val
elif markup_type == "GC":
tag, val = try_split(1)
tag = tag.replace("_", " ")
file_markups[tag] = file_markups.get(tag, "") + val
elif markup_type == "GR":
seq_name, tag, val = try_split(2)
tag = tag.replace("_", " ")
seq_markups.setdefault(seq_name, {}).setdefault(tag, "")
seq_markups[seq_name][tag] += val
# ignore other types
continue
elif line.startswith('#'):
# unstructured comment
if 'comments' in file_attrs:
file_attrs['comments'] += "\n" + line[1:]
else:
file_attrs['comments'] = line[1:]
continue
elif line.strip() == "//":
# end of sequence alignment blocks, but comments may follow this, so keep going...
continue
# sequence info...
try:
seq_name, block = line.split(None, 1)
except ValueError:
raise FormatSyntaxError(
"Sequence info not in name/contents format on line %d" %
line_num)
if seq_name not in sequences:
sequences[seq_name] = Sequence(name=make_readable(seq_name))
seq_sequence.append(seq_name)
sequences[seq_name].extend(block)
f.close()
for seq_name, seq in sequences.items():
if seq_name in seq_attrs:
seq.attrs = seq_attrs[seq_name]
if seq_name in seq_markups:
seq.markups = seq_markups[seq_name]
for tag, markup in seq.markups.items():
if len(markup) != len(seq):
session.logger.warning(
"Markup %s for sequence %s is wrong length; ignoring" %
(tag, seq_name))
del seq.markups[tag]
for seq_info, label in [(seq_attrs, "sequence"), (seq_markups, "residue")]:
for seq_name in seq_info.keys():
if seq_name in sequences:
continue
# might be sequence name if trailing '/start-end' is removed...
for full_name in sequences.keys():
if full_name.startswith(seq_name) \
and full_name[len(seq_name)] == '/' \
and '/' not in full_name[len(seq_name)+1:]:
break
else:
raise FormatSyntaxError(
"%s annotations provided for non-existent sequence %s" %
(label.capitalize(), seq_name))
session.logger.info(
"Updating %s %s annotations with %s annotations" %
(full_name, label, seq_name))
seq_info[full_name].update(seq_info[seq_name])
del seq_info[seq_name]
for tag, markup in file_markups.items():
if len(markup) != len(sequences[seq_sequence[0]]):
raise FormatSyntaxError("Column annotation %s is wrong length" %
tag)
return [sequences[name] for name in seq_sequence], file_attrs, file_markups
def read(session, f):
IN_HEADER, START_ATTRS, IN_ATTRS, IN_FEATURES, IN_SEQ = range(5)
state = IN_HEADER
sequences = []
line_num = 0
has_offset = False
longest = None
file_attrs = {}
for line in f.readlines():
line = line.rstrip() # remove trailing whitespace/newline
line_num += 1
if line_num == 1:
if line.startswith("!!RICH_SEQUENCE"):
continue
raise FormatSyntaxError("First line does not start with !!RICH_SEQUENCE")
if state == IN_HEADER:
if line.strip() == "..":
state = START_ATTRS
continue
if "comments" in file_attrs:
file_attrs["comments"] += "\n" + line
else:
file_attrs["comments"] = line
continue
if not line.strip():
continue
if state == START_ATTRS:
if line.strip() == "{":
state = IN_ATTRS
cur_attr = None
attrs = {}
elif line:
raise FormatSyntaxError(
"Unexpected text before start of sequence on line %d" &line_num)
continue
if state == IN_ATTRS or state == IN_FEATURES:
if line.strip() == "sequence" and line[0] == "s":
if "RSF name" not in attrs:
raise FormatSyntaxError("Sequence on line %d has no name" & line_num)
state = IN_SEQ
seq = Sequence(name=make_readable(attrs["RSF name"]))
del attrs["RSF name"]
seq.attrs = attrs
if "RSF descrip" in attrs:
attrs["description"] = attrs["RSF descrip"]
del attrs["RSF descrip"]
sequences.append(seq)
if "RSF offset" in attrs:
seq.extend("." * int(attrs["RSF offset"]))
has_offset = True
del attrs["RSF offset"]
continue
if line.startswith("feature"):
if state == IN_ATTRS:
attrs["RSF features"] = [[line[8:]]]
else:
attrs["RSF features"].append([line[8:]])
state = IN_FEATURES
continue
if state == IN_ATTRS:
if line[0].isspace():
# continuation
if not cur_attr:
raise FormatSyntaxError("Bogus indentation at line %d" % line_num)
if attrs[cur_attr]:
attrs[cur_attr] += "\n" + line
else:
attrs[cur_attr] = line
continue
if " " in line.strip():
cur_attr, val = line.split(None, 1)
cur_attr.replace("_", " ")
cur_attr = "RSF " + cur_attr
attrs[cur_attr] = val.strip()
else:
cur_attr = "RSF " + line.strip().replace("_", " ")
attrs[cur_attr] = ""
continue
if state == IN_FEATURES:
attrs["RSF features"][-1].append(line)
continue
if line.strip() == "}":
state = START_ATTRS
if not longest:
longest = len(seq)
else:
if len(seq) < longest:
seq.extend("." * (longest - len(seq)))
elif len(seq) > longest:
longest = len(seq)
for s in sequences[:-1]:
s.extend("." * (longest - len(s)))
continue
seq.extend(line.strip())
if not seq[0].isalpha():
has_offset = True
f.close()
if state == IN_HEADER:
raise FormatSyntaxError("No end to header (i.e. '..' line) found")
if state == IN_ATTRS or state == IN_FEATURES:
raise FormatSyntaxError("No sequence data found for sequence %s" % attrs["RSF name"])
if state == IN_SEQ:
raise FormatSyntaxError("No terminating brace for sequence %s" % attrs["RSF name"])
if not has_offset:
session.logger.warning("No offset fields in RSF file; assuming zero offset")
return sequences, file_attrs, {}
def _update_errors_gaps(self, aseq):
if not self.settings.error_region_shown and not self.settings.gap_region_shown:
return
a_ref_seq = getattr(aseq, 'residue_sequence', aseq.ungapped())
errors = [0] * len(a_ref_seq)
gaps = [0] * len(a_ref_seq)
from chimerax.atomic import Sequence
for chain, match_map in aseq.match_maps.items():
for i, char in enumerate(a_ref_seq):
try:
res = match_map[i]
except KeyError:
gaps[i] += 1
else:
if Sequence.rname3to1(res.name) != char.upper():
errors[i] += 1
partial_error_blocks, full_error_blocks = [], []
partial_gap_blocks, full_gap_blocks = [], []
num_assocs = len(aseq.match_maps)
if num_assocs > 0:
for partial, full, check in [
(partial_error_blocks, full_error_blocks, errors),
(partial_gap_blocks, full_gap_blocks, gaps)
]:
cur_partial_block = cur_full_block = None
for i, check_num in enumerate(check):
gapped_i = aseq.ungapped_to_gapped(i)
if check_num == num_assocs:
if cur_full_block:
cur_full_block[-1] = gapped_i
else:
cur_full_block = [aseq, aseq, gapped_i, gapped_i]
full.append(cur_full_block)
if cur_partial_block:
cur_partial_block = None
else:
if cur_full_block:
cur_full_block = None
if check_num > 0:
if cur_partial_block:
cur_partial_block[-1] = gapped_i
else:
cur_partial_block = [
aseq, aseq, gapped_i, gapped_i
]
partial.append(cur_partial_block)
elif cur_partial_block:
cur_partial_block = None
for shown, region_name_part, partial_blocks, full_blocks, fills, outlines in [
(self.settings.error_region_shown, self.ERROR_REGION_STRING,
partial_error_blocks, full_error_blocks,
self.settings.error_region_interiors,
self.settings.error_region_borders),
(self.settings.gap_region_shown, self.GAP_REGION_STRING,
partial_gap_blocks, full_gap_blocks,
self.settings.gap_region_interiors,
self.settings.gap_region_borders)
]:
if not shown:
continue
full_fill, partial_fill = fills
full_outline, partial_outline = outlines
for region_name_start, blocks, fill, outline in [
(region_name_part, full_blocks, full_fill, full_outline),
("partial " + region_name_part, partial_blocks, partial_fill,
partial_outline)
]:
region_name = "%s of %s" % (region_name_start, aseq.name)
old_reg = self.region_browser.get_region(region_name,
create=False)
if old_reg:
self.region_browser.delete_region(old_reg)
if blocks:
self.region_browser.new_region(region_name,
blocks=blocks,
fill=fill,
outline=outline,
sequence=aseq,
cover_gaps=False)
def model(session,
targets,
*,
block=True,
multichain=True,
custom_script=None,
dist_restraints=None,
executable_location=None,
fast=False,
het_preserve=False,
hydrogens=False,
license_key=None,
num_models=5,
show_gui=True,
temp_path=None,
thorough_opt=False,
water_preserve=False):
"""
Generate comparative models for the target sequences.
Arguments:
session
current session
targets
list of (alignment, sequence) tuples. Each sequence will be modelled.
block
If True, wait for modelling job to finish before returning and return list of
(opened) models. Otherwise return immediately. Also see 'show_gui' option.
multichain
If True, the associated chains of each structure are used individually to generate
chains in the resulting models (i.e. the models will be multimers). If False, all
associated chains are used together as templates to generate a single-chain model
for the target sequence.
custom_script
If provided, the location of a custom Modeller script to use instead of the
one we would otherwise generate. Only used when executing locally.
dist_restraints
If provided, the location of a file containing additional distance restraints
executable_location
If provided, the path to the locally installed Modeller executable. If not
provided, use the web service.
fast
Whether to use fast but crude generation of models
het_preserve
Whether to preserve HET atoms in generated models
hydrogens
Whether to generate models with hydrogen atoms
license_key
Modeller license key. If not provided, try to use settings to find one.
num_models
Number of models to generate for each template sequence
show_gui
If True, show user interface for Modeller results (if ChimeraX is in gui mode).
temp_path
If provided, folder to use for temporary files
thorough_opt
Whether to perform thorough optimization
water_preserve
Whether to preserve water in generated models
"""
from chimerax.core.errors import LimitationError, UserError
from .common import modeller_copy
if multichain:
# So, first find structure with most associated chains and least non-associated chains.
# That structure is used as the multimer template. Chains from other structures are used
# as "standalone" templates -- each such chain will be on its own line. Need to allow
# space on the left and right of the target sequence so that the largest chains can be
# accomodated.
# Find the structure we will use as the multimer template
by_structure = {}
chain_info = {}
for alignment, orig_target in targets:
# Copy the target sequence, changing name to conform to Modeller limitations
target = modeller_copy(orig_target)
if not alignment.associations:
raise UserError("Alignment %s has no associated chains" %
alignment.ident)
for chain, aseq in alignment.associations.items():
if len(chain.chain_id) > 1:
raise LimitationError(
"Modeller cannot handle templates with multi-character chain IDs"
)
by_structure.setdefault(chain.structure, []).append(chain)
chain_info[chain] = (aseq, target)
max_matched = min_unmatched = None
for s, match_info in by_structure.items():
matched = len(match_info)
unmatched = s.num_chains - len(match_info)
if max_matched is None or matched > max_matched or (
matched == max_matched and (unmatched < min_unmatched)):
multimer_template = s
max_matched = matched
min_unmatched = unmatched
mm_targets = []
mm_chains = []
match_chains = []
for chain in multimer_template.chains:
mm_chains.append(chain)
try:
aseq, target = chain_info[chain]
except KeyError:
mm_targets.append(None)
else:
mm_targets.append(target)
match_chains.append(chain)
# okay, now form single-chain lines for the other structure associations, that eventually will
# be handled column by column in exactly the same way as the non-multichain method.
single_template_lines = []
for chain, info in chain_info.items():
if chain.structure == multimer_template:
continue
aseq, target = info
for i, mm_target in enumerate(mm_targets):
if mm_target != target:
continue
template_line = [None] * len(mm_targets)
template_line[i] = chain
single_template_lines.append(template_line)
# AFAIK, the multimer template chain sequences need to have complete PDB sequence, so may need
# to prefix and suffix he corresponding alignment sequence with characters for residues
# outside of the alignment sequence. For other templates/targets, affix a corresponding number
# of '-' characters
prefixes, suffixes = find_affixes(mm_chains, chain_info)
target_strings = []
for prefix, suffix, mm_target in zip(prefixes, suffixes, mm_targets):
if mm_target is None:
target_strings.append('-')
continue
target_strings.append('-' * len(prefix) + mm_target.characters +
'-' * len(suffix))
templates_strings = []
templates_info = []
mm_template_strings = []
for prefix, suffix, chain in zip(prefixes, suffixes, mm_chains):
try:
aseq, target = chain_info[chain]
except KeyError:
mm_template_strings.append('-')
continue
mm_template_strings.append(
prefix + regularized_seq(aseq, chain).characters + suffix)
templates_strings.append(mm_template_strings)
templates_info.append(None)
for template_line in single_template_lines:
template_strings = []
for prefix, suffix, chain, target in zip(prefixes, suffixes,
template_line,
mm_targets):
if target is None:
template_strings.append('-')
elif chain is None:
template_strings.append(
'-' * (len(prefix) + len(target) + len(suffix)))
else:
aseq, target = chain_info[chain]
template_strings.append(
'-' * len(prefix) +
regularized_seq(aseq, chain).characters +
'-' * len(suffix))
templates_info.append((chain, aseq.match_maps[chain]))
templates_strings.append(template_strings)
target_name = "target" if len(targets) > 1 else target.name
else:
if len(targets) > 1:
raise LimitationError(
"Cannot have multiple targets(/alignments) unless creating multimeric model"
)
alignment, orig_target = targets[0]
# Copy the target sequence, changing name to conform to Modeller limitations
target = modeller_copy(orig_target)
target_strings = [target.characters]
templates_strings = []
templates_info = []
match_chains = []
for chain, aseq in alignment.associations.items():
if len(chain.chain_id) > 1:
raise LimitationError(
"Modeller cannot handle templates with multi-character chain IDs"
)
templates_strings.append([regularized_seq(aseq, chain).characters])
templates_info.append((chain, aseq.match_maps[chain]))
if not match_chains:
match_chains.append(chain)
target_name = target.name
from .common import write_modeller_scripts, get_license_key
script_path, config_path, temp_dir = write_modeller_scripts(
get_license_key(session, license_key), num_models, het_preserve,
water_preserve, hydrogens, fast, None, custom_script, temp_path,
thorough_opt, dist_restraints)
input_file_map = []
# form the sequences to be written out as a PIR
from chimerax.atomic import Sequence
pir_target = Sequence(name=target_name)
pir_target.description = "sequence:%s:.:.:.:.::::" % pir_target.name
pir_target.characters = '/'.join(target_strings)
pir_seqs = [pir_target]
structures_to_save = set()
for strings, info in zip(templates_strings, templates_info):
if info is None:
# multimer template
pir_template = Sequence(
name=structure_save_name(multimer_template))
pir_template.description = "structure:%s:FIRST:%s::::::" % (
pir_template.name, multimer_template.chains[0].chain_id)
structures_to_save.add(multimer_template)
else:
# single-chain template
chain, match_map = info
first_assoc_pos = 0
while first_assoc_pos not in match_map:
first_assoc_pos += 1
first_assoc_res = match_map[first_assoc_pos]
pir_template = Sequence(name=chain_save_name(chain))
pir_template.description = "structure:%s:%d%s:%s:+%d:%s::::" % (
structure_save_name(chain.structure), first_assoc_res.number,
first_assoc_res.insertion_code, chain.chain_id, len(match_map),
chain.chain_id)
structures_to_save.add(chain.structure)
pir_template.characters = '/'.join(strings)
pir_seqs.append(pir_template)
import os.path
pir_file = os.path.join(temp_dir.name, "alignment.ali")
aln = session.alignments.new_alignment(pir_seqs,
False,
auto_associate=False,
create_headers=False)
aln.save(pir_file, format_name="pir")
session.alignments.destroy_alignment(aln)
input_file_map.append(("alignment.ali", "text_file", pir_file))
# write the namelist.dat file, target seq name on first line, templates on remaining lines
name_file = os.path.join(temp_dir.name, "namelist.dat")
input_file_map.append(("namelist.dat", "text_file", name_file))
with open(name_file, 'w') as f:
for template_seq in pir_seqs:
print(template_seq.name, file=f)
config_name = os.path.basename(config_path)
input_file_map.append((config_name, "text_file", config_path))
# save structure files
import os
struct_dir = os.path.join(temp_dir.name, "template_struc")
if not os.path.exists(struct_dir):
try:
os.mkdir(struct_dir, mode=0o755)
except FileExistsError:
pass
from chimerax.pdb import save_pdb, standard_polymeric_res_names as std_res_names
for structure in structures_to_save:
base_name = structure_save_name(structure) + '.pdb'
pdb_file_name = os.path.join(struct_dir, base_name)
input_file_map.append((base_name, "text_file", pdb_file_name))
ATOM_res_names = structure.in_seq_hets
ATOM_res_names.update(std_res_names)
save_pdb(session,
pdb_file_name,
models=[structure],
polymeric_res_names=ATOM_res_names)
delattr(structure, 'in_seq_hets')
from chimerax.atomic import Chains
match_chains = Chains(match_chains)
if executable_location is None:
if custom_script is not None:
raise LimitationError(
"Custom Modeller scripts only supported when executing locally"
)
if dist_restraints is not None:
raise LimitationError(
"Distance restraints only supported when executing locally")
if thorough_opt:
session.logger.warning(
"Thorough optimization only supported when executing locally")
job_runner = ModellerWebService(session, match_chains, num_models,
pir_target.name, input_file_map,
config_name, targets, show_gui)
else:
#TODO: job_runner = ModellerLocal(...)
from chimerax.core.errors import LimitationError
raise LimitationError("Local Modeller execution not yet implemented")
# a custom script [only used when executing locally] needs to be copied into the tmp dir...
if os.path.exists(script_path) \
and os.path.normpath(temp_dir.name) != os.path.normpath(os.path.dirname(script_path)):
import shutil
shutil.copy(script_path, temp_dir.name)
return job_runner.run(block=block)
def _prep_add(session,
structures,
unknowns_info,
template,
need_all=False,
**prot_schemes):
global _serial
_serial = None
atoms = []
type_info_for_atom = {}
naming_schemas = {}
idatm_type = {} # need this later; don't want a recomp
hydrogen_totals = {}
# add missing OXTs of "real" C termini;
# delete hydrogens of "fake" N termini after protonation
# and add a single "HN" back on, using same dihedral as preceding residue;
# delete extra hydrogen of "fake" C termini after protonation
logger = session.logger
real_N, real_C, fake_N, fake_C = determine_termini(session, structures)
logger.info("Chain-initial residues that are actual N"
" termini: %s" % ", ".join([str(r) for r in real_N]))
logger.info("Chain-initial residues that are not actual N"
" termini: %s" % ", ".join([str(r) for r in fake_N]))
logger.info("Chain-final residues that are actual C"
" termini: %s" % ", ".join([str(r) for r in real_C]))
logger.info("Chain-final residues that are not actual C"
" termini: %s" % ", ".join([str(r) for r in fake_C]))
for rc in real_C:
complete_terminal_carboxylate(session, rc)
# ensure that N termini are protonated as N3+ (since Npl will fail)
from chimerax.atomic import Sequence
for nter in real_N + fake_N:
n = nter.find_atom("N")
if not n:
continue
# if residue wasn't templated, leave atom typing alone
if Sequence.protein3to1(n.residue.name) == 'X':
continue
if not (n.residue.name == "PRO" and n.num_bonds >= 2):
n.idatm_type = "N3+"
coordinations = {}
for struct in structures:
pbg = struct.pseudobond_group(struct.PBG_METAL_COORDINATION,
create_type=None)
if not pbg:
continue
for pb in pbg.pseudobonds:
for a in pb.atoms:
if not need_all and a.structure not in structures:
continue
if not a.element.is_metal:
coordinations.setdefault(a, []).append(pb.other_atom(a))
remaining_unknowns = {}
type_info_class = type_info['H'].__class__
from chimerax.atomic import Residue
for struct in structures:
for atom in struct.atoms:
if atom.element.number == 0:
res = atom.residue
struct.delete_atom(atom)
idatm_lookup = {}
if template:
template_lookup = {}
from chimerax.atomic import TmplResidue
get_template = TmplResidue.get_template
for res in struct.residues:
if get_template(res.name):
continue
try:
exemplar = template_lookup[res.name]
except KeyError:
from chimerax.mmcif import find_template_residue
tmpl = find_template_residue(session, res.name)
if not tmpl:
continue
from chimerax.atomic import AtomicStructure
s = AtomicStructure(session)
r = exemplar = template_lookup[res.name] = s.new_residue(
res.name, 'A', 1)
atom_map = {}
for ta in tmpl.atoms:
if ta.element.number > 1:
a = s.new_atom(ta.name, ta.element)
a.coord = ta.coord
r.add_atom(a)
atom_map[ta] = a
for tnb in ta.neighbors:
if tnb in atom_map:
s.new_bond(a, atom_map[tnb])
for a in res.atoms:
ea = exemplar.find_atom(a.name)
if ea:
a.idatm_type = ea.idatm_type
for r in template_lookup.values():
r.structure.delete()
template_lookup.clear()
for atom in struct.atoms:
atom_type = atom.idatm_type
idatm_type[atom] = atom_type
if atom_type in type_info:
# don't want to ask for idatm_type in middle
# of hydrogen-adding loop (since that will
# force a recomp), so remember here
type_info_for_atom[atom] = type_info[atom_type]
# if atom is in standard residue but has missing bonds to
# heavy atoms, skip it instead of incorrectly protonating
# (or possibly throwing an error if e.g. it's planar)
# also
# UNK/N residues will be missing some or all of their side-chain atoms, so
# skip atoms that would otherwise be incorrectly protonated due to their
# missing neighbors
truncated = \
atom.is_missing_heavy_template_neighbors(no_template_okay=True) \
or \
(atom.residue.name in ["UNK", "N"] and atom.residue.polymer_type != Residue.PT_NONE
and unk_atom_truncated(atom)) \
or \
(atom.residue.polymer_type == Residue.PT_NUCLEIC and atom.name == "P"
and atom.num_explicit_bonds < 4)
if truncated:
session.logger.warning(
"Not adding hydrogens to %s because it is missing heavy-atom"
" bond partners" % atom)
type_info_for_atom[atom] = type_info_class(
4, atom.num_bonds, atom.name)
else:
atoms.append(atom)
# sulfonamide nitrogens coordinating a metal
# get an additional hydrogen stripped
if coordinations.get(atom, []) and atom.element.name == "N":
if "Son" in [nb.idatm_type for nb in atom.neighbors]:
orig_ti = type_info[atom_type]
type_info_for_atom[atom] = orig_ti.__class__(
orig_ti.geometry, orig_ti.substituents - 1,
orig_ti.description)
continue
if atom in unknowns_info:
type_info_for_atom[atom] = unknowns_info[atom]
atoms.append(atom)
continue
remaining_unknowns.setdefault(atom.residue.name,
set()).add(atom.name)
# leave remaining unknown atoms alone
type_info_for_atom[atom] = type_info_class(4, atom.num_bonds,
atom.name)
for rname, atom_names in remaining_unknowns.items():
names_text = ", ".join([nm for nm in atom_names])
atom_text, obj_text = ("atoms",
"them") if len(atom_names) > 1 else ("atom",
"it")
logger.warning(
"Unknown hybridization for %s (%s) of residue type %s;"
" not adding hydrogens to %s" %
(atom_text, names_text, rname, obj_text))
naming_schemas.update(
determine_naming_schemas(struct, type_info_for_atom))
if need_all:
from chimerax.atomic import AtomicStructure
for struct in [
m for m in session.models if isinstance(m, AtomicStructure)
]:
if struct in structures:
continue
for atom in struct.atoms:
idatm_type[atom] = atom.idatm_type
if atom.idatm_type in type_info:
type_info_for_atom[atom] = type_info[atom.idatm_type]
for atom in atoms:
if atom not in type_info_for_atom:
continue
bonding_info = type_info_for_atom[atom]
total_hydrogens = bonding_info.substituents - atom.num_bonds
for bonded in atom.neighbors:
if bonded.element.number == 1:
total_hydrogens += 1
hydrogen_totals[atom] = total_hydrogens
schemes = {}
# HIS and CYS treated as 'unspecified'; use built-in typing
for scheme_type, res_names, res_check, typed_atoms in [
('his', ["HID", "HIE", "HIP"], None, []),
('asp', asp_res_names, _asp_check, asp_prot_names),
('glu', glu_res_names, _glu_check, glu_prot_names),
('lys', ["LYS", "LYN"], _lys_check, ["NZ"]),
('cys', ["CYM"], _cys_check, ["SG"])
]:
scheme = prot_schemes.get(scheme_type + '_scheme', None)
if scheme is None:
by_name = True
scheme = {}
else:
by_name = False
if not scheme:
for s in structures:
for r in s.residues:
if r.name in res_names and res_check and res_check(r):
if by_name:
scheme[r] = r.name
elif scheme_type != 'his':
scheme[r] = res_names[0]
# unset any explicit typing...
for ta in typed_atoms:
a = r.find_atom(ta)
if a:
a.idatm_type = None
else:
for r in scheme.keys():
if res_check and not res_check(r, scheme[r]):
del scheme[r]
schemes[scheme_type] = scheme
# create dictionary keyed on histidine residue with value of another
# dictionary keyed on the nitrogen atoms with boolean values: True
# equals should be protonated
his_Ns = {}
for r, protonation in schemes["his"].items():
delta = r.find_atom("ND1")
epsilon = r.find_atom("NE2")
if delta is None or epsilon is None:
# find the ring, etc.
rings = r.structure.rings()
for ring in rings:
if r in rings.atoms.residues:
break
else:
continue
# find CG by locating CB-CG bond
ring_bonds = ring.bonds
for ra in ring.atoms:
if ra.element.name != "C":
continue
for ba, b in zip(ra.neighbors, ra.bonds):
if ba.element.name == "C" and b not in ring_bonds:
break
else:
continue
break
else:
continue
nitrogens = [a for a in ring.atoms if a.element.name == "N"]
if len(nitrogens) != 2:
continue
if ra in nitrogens[0].neighbors:
delta, epsilon = nitrogens
else:
epsilon, delta = nitrogens
if protonation == "HID":
his_Ns.update({delta: True, epsilon: False})
elif protonation == "HIE":
his_Ns.update({delta: False, epsilon: True})
elif protonation == "HIP":
his_Ns.update({delta: True, epsilon: True})
else:
continue
for n, do_prot in his_Ns.items():
if do_prot:
type_info_for_atom[n] = type_info["Npl"]
n.idatm_type = idatm_type[n] = "Npl"
else:
type_info_for_atom[n] = type_info["N2"]
n.idatm_type = idatm_type[n] = "N2"
for r, protonation in schemes["asp"].items():
_handle_acid_protonation_scheme_item(r, protonation, asp_res_names,
asp_prot_names, type_info,
type_info_for_atom)
for r, protonation in schemes["glu"].items():
_handle_acid_protonation_scheme_item(r, protonation, glu_res_names,
glu_prot_names, type_info,
type_info_for_atom)
for r, protonation in schemes["lys"].items():
nz = r.find_atom("NZ")
if protonation == "LYS":
it = 'N3+'
else:
it = 'N3'
ti = type_info[it]
if nz is not None:
type_info_for_atom[nz] = ti
# avoid explicitly setting type if possible
if nz.idatm_type != it:
nz.idatm_type = it
for r, protonation in schemes["cys"].items():
sg = r.find_atom("SG")
if protonation == "CYS":
it = 'S3'
else:
it = 'S3-'
ti = type_info[it]
if sg is not None:
type_info_for_atom[sg] = ti
# avoid explicitly setting type if possible
if sg.idatm_type != it:
sg.idatm_type = it
return atoms, type_info_for_atom, naming_schemas, idatm_type, \
hydrogen_totals, his_Ns, coordinations, fake_N, fake_C
def align(session,
ref,
match,
matrix_name,
algorithm,
gap_open,
gap_extend,
dssp_cache,
ss_matrix=defaults["ss_scores"],
ss_fraction=defaults["ss_mixture"],
gap_open_helix=defaults["helix_open"],
gap_open_strand=defaults["strand_open"],
gap_open_other=defaults["other_open"],
compute_ss=defaults["compute_ss"]):
from chimerax import sim_matrices
similarity_matrix = sim_matrices.matrix(matrix_name, session.logger)
ssf = ss_fraction
ssm = ss_matrix
if ssf is not None and ssf is not False and compute_ss:
need_compute = []
if ref.structure not in dssp_cache:
for r in ref.residues:
if r and len(r.atoms) > 1:
# not CA only
need_compute.append(ref.structure)
dssp_cache[ref.structure] = (
ref.structure.residues.ss_ids,
ref.structure.residues.ss_types)
break
if match.structure not in dssp_cache:
for r in match.residues:
if r and len(r.atoms) > 1:
# not CA only
need_compute.append(match.structure)
dssp_cache[match.structure] = (
match.structure.residues.ss_ids,
match.structure.residues.ss_types)
break
if need_compute:
"""TODO
from chimera.initprefs import ksdsspPrefs, \
KSDSSP_ENERGY, KSDSSP_HELIX_LENGTH, \
KSDSSP_STRAND_LENGTH
"""
from chimerax.std_commands import dssp
dssp.compute_ss(session, need_compute)
if algorithm == "nw":
from chimerax.alignment_algs import NeedlemanWunsch
score, seqs = NeedlemanWunsch.nw(ref,
match,
score_gap=-gap_extend,
score_gap_open=0 - gap_open,
similarity_matrix=similarity_matrix,
return_seqs=True,
ss_matrix=ss_matrix,
ss_fraction=ss_fraction,
gap_open_helix=-gap_open_helix,
gap_open_strand=-gap_open_strand,
gap_open_other=-gap_open_other)
gapped_ref, gapped_match = seqs
elif algorithm == "sw":
def ss_let(r):
if not r:
return ' '
if r.is_helix:
return 'H'
elif r.is_strand:
return 'S'
return 'O'
if ssf is False or ssf is None:
ssf = 0.0
ssm = None
if ssm:
# account for missing structure (blank SS letter)
ssm = ssm.copy()
for let in "HSO ":
ssm[(let, ' ')] = 0.0
ssm[(' ', let)] = 0.0
from chimerax.alignment_algs import SmithWaterman
score, alignment = SmithWaterman.align(
ref.characters,
match.characters,
similarity_matrix,
float(gap_open),
float(gap_extend),
gap_char=".",
ss_matrix=ssm,
ss_fraction=ssf,
gap_open_helix=float(gap_open_helix),
gap_open_strand=float(gap_open_strand),
gap_open_other=float(gap_open_other),
ss1="".join([ss_let(r) for r in ref.residues]),
ss2="".join([ss_let(r) for r in match.residues]))
from chimerax.atomic import StructureSeq, Sequence
gapped_ref = StructureSeq(structure=ref.structure,
chain_id=ref.chain_id)
gapped_ref.name = ref.structure.name
gapped_match = StructureSeq(structure=match.structure,
chain_id=match.chain_id)
gapped_match.name = match.structure.name
# Smith-Waterman may not be entirety of sequences...
for orig, gapped, sw in [
(ref, gapped_ref, Sequence(characters=alignment[0])),
(match, gapped_match, Sequence(characters=alignment[1]))
]:
ungapped = sw.ungapped()
for i in range(len(orig) - len(ungapped) + 1):
if ungapped == orig[i:i + len(ungapped)]:
break
else:
raise ValueError("Smith-Waterman result not"
" a subsequence of original sequence")
gapped.bulk_set(orig.residues[i:i + len(ungapped)], sw.characters)
else:
raise ValueError("Unknown sequence alignment algorithm: %s" %
algorithm)
# If the structures are disjoint snippets of the same longer SEQRES,
# they may be able to be structurally aligned but the SEQRES records
# will keep them apart. Try to detect this situation and work around
# by snipping off sequence ends.
sr_disjoint = False
if ref.from_seqres and match.from_seqres:
struct_match = 0
for i in range(len(gapped_ref)):
uri = gapped_ref.gapped_to_ungapped(i)
if uri is None:
continue
umi = gapped_match.gapped_to_ungapped(i)
if umi is None:
continue
if gapped_ref.residues[uri] and gapped_match.residues[umi]:
struct_match += 1
if struct_match >= 3:
break
if struct_match < 3:
seq_match = 0
for s1, s2 in zip(gapped_ref[:], gapped_match[:]):
if s1.isalpha() and s2.isalpha():
seq_match += 1
if seq_match > 3:
break
if seq_match > 3:
need = 3 - struct_match
if (ref.residues[:need].count(None) == 3
or ref.residues[-need:].count(None) == 3) \
and (match.residues[:need].count(None) == 3
or match.residues[-need:].count(None) == 3):
sr_disjoint = True
if sr_disjoint:
from copy import copy
clipped_ref = copy(ref)
clipped_match = copy(match)
for seq in (clipped_ref, clipped_match):
num_none = 0
for r in seq.residues:
if r:
break
num_none += 1
if num_none:
seq.bulk_set(seq.residues[num_none:], seq[num_none:])
num_none = 0
for r in reversed(seq.residues):
if r:
break
num_none += 1
if num_none:
seq.bulk_set(seq.residues[:-num_none], seq[:-num_none])
return align(session,
clipped_ref,
clipped_match,
matrix_name,
algorithm,
gap_open,
gap_extend,
dssp_cache,
ss_matrix=ss_matrix,
ss_fraction=ss_fraction,
gap_open_helix=gap_open_helix,
gap_open_strand=gap_open_strand,
gap_open_other=gap_open_other,
compute_ss=False)
for orig, aligned in [(ref, gapped_ref), (match, gapped_match)]:
if hasattr(orig, '_dm_rebuild_info'):
aligned._dm_rebuild_info = orig._dm_rebuild_info
_dm_cleanup.append(aligned)
return score, gapped_ref, gapped_match
def write_mol2(session,
file_name,
*,
models=None,
atoms=None,
status=None,
anchor=None,
rel_model=None,
sybyl_hyd_naming=True,
combine_models=False,
skip_atoms=None,
res_num=False,
gaff_type=False,
gaff_fail_error=None):
"""Write a Mol2 file.
Parameters
----------
file_name : str, or file object open for writing
Output file.
models : a list/tuple/set of models (:py:class:`~chimerax.atomic.Structure`s)
or a single :py:class:`~chimerax.atomic.Structure`
The structure(s) to write out. If None (and 'atoms' is also None) then
write out all structures.
atoms : an :py:class:`~chimerax.atomic.Atoms` collection or None. If not None,
then 'models' must be None.
status : function or None
If not None, a function that takes a string -- used to report the progress of the write.
anchor : :py:class:`~chimerax.atomic.Atoms` collection
Atoms (and their implied internal bonds) that should be written out to the
@SET section of the file as the rigid framework for flexible ligand docking.
rel_model : Model whose coordinate system the coordinates should be written out reletive to,
i.e. take the output atoms' coordinates and apply the inverse of the rel_model's transform.
sybyl_hyd_naming : bool
Controls whether hydrogen names should be "Sybyl-like" or "PDB-like" -- e.g. HG21 vs. 1HG2.
combine_models : bool
Controls whether multiple structures will be combined into a single @MOLECULE
section (value: True) or each given its own section (value: False).
skip_atoms : list/set of :py:class:`~chimerax.atomic.Atom`s or an :py:class:`~chimerax.atomic.Atoms` collection or None
Atoms to not output
res_num : bool
Controls whether residue sequence numbers are included in the substructure name.
Since Sybyl Mol2 files include them, this defaults to True.
gaff_type : bool
If 'gaff_type' is True, outout GAFF atom types instead of Sybyl atom types.
`gaff_fail_error`, if specified, is the type of error to throw (e.g. UserError)
if there is no gaff_type attribute for an atom, otherwise throw the standard AttributeError.
"""
if status:
status("Writing Mol2 file %s" % file_name)
from chimerax import io
f = io.open_output(file_name, "utf-8")
sort_key_func = serial_sort_key = lambda a, ri={}: write_mol2_sort_key(
a, res_indices=ri)
from chimerax.atomic import Structure, Atoms, Residue
class JPBGroup:
def __init__(self, atoms):
atom_set = set(atoms)
pbs = []
for s in atoms.unique_structures:
pbg = s.pbg_map.get(s.PBG_METAL_COORDINATION, None)
if not pbg:
continue
for pb in pbg.pseudobonds:
if pb.atoms[0] in atom_set and pb.atoms[1] in atom_set:
pbs.append(pb)
self._pbs = pbs
@property
def pseudobonds(self):
return self._pbs
if models is None:
if atoms is None:
structures = session.models.list(type=Structure)
else:
structures = atoms
else:
if atoms is None:
if isinstance(models, Structure):
structures = [models]
else:
structures = [m for m in models if isinstance(m, Structure)]
else:
raise ValueError(
"Cannot specify both 'models' and 'atoms' keywords")
if isinstance(structures, Atoms):
class Jumbo:
def __init__(self, atoms):
self.atoms = atoms
self.residues = atoms.unique_residues
self.bonds = atoms.intra_bonds
self.name = "(selection)"
self.pbg_map = {
Structure.PBG_METAL_COORDINATION: JPBGroup(atoms)
}
structures = [Jumbo(structures)]
sort_key_func = lambda a: (a.structure.id, ) + serial_sort_key(a)
combine_models = False
# transform...
if rel_model is None:
from chimerax.geometry import identity
xform = identity()
else:
xform = rel_model.scene_position.inverse()
# need to find amide moieties since Sybyl has an explicit amide type
if status:
status("Finding amides")
from chimerax.chem_group import find_group
amides = find_group("amide", structures)
amide_Ns = set([amide[2] for amide in amides])
amide_CNs = set([amide[0] for amide in amides])
amide_CNs.update(amide_Ns)
amide_Os = set([amide[1] for amide in amides])
substructure_names = None
if combine_models and len(structures) > 1:
# create a fictitious jumbo model
class Jumbo:
def __init__(self, structures):
self.name = structures[0].name + " (combined)"
from chimerax.atomic import concatenate
self.atoms = concatenate([s.atoms for s in structures])
self.bonds = concatenate([s.bonds for s in structures])
self.residues = concatenate([s.residues for s in structures])
self.pbg_map = {
Structure.PBG_METAL_COORDINATION: JPBGroup(self.atoms)
}
# if combining single-residue structures,
# can be more informative to use model name
# instead of residue type for substructure
if len(structures) == len(self.residues):
rnames = self.residues.names
if len(set(rnames)) < len(rnames):
snames = [s.name for s in structures]
if len(set(snames)) == len(snames):
self.substructure_names = dict(
zip(self.residues, snames))
structures = [Jumbo(structures)]
if hasattr(structures[-1], 'substructure_names'):
substructure_names = structures[-1].substructure_names
delattr(structures[-1], 'substructure_names')
sort_key_func = lambda a: (a.structure.id, ) + serial_sort(a)
# write out structures
for struct in structures:
if hasattr(struct, 'mol2_comments'):
for m2c in struct.mol2_comments:
print(m2c, file=f)
if hasattr(struct, 'solvent_info'):
print(struct.solvent_info, file=f)
# molecule section header
print("%s" % MOLECULE_HEADER, file=f)
# molecule name
print("%s" % struct.name, file=f)
atoms = list(struct.atoms)
bonds = list(struct.bonds)
# add metal-coordination bonds
coord_grp = struct.pbg_map.get(Structure.PBG_METAL_COORDINATION, None)
if coord_grp:
bonds.extend(list(coord_grp.pseudobonds))
if skip_atoms:
skip_atoms = set(skip_atoms)
atoms = [a for a in atoms if a not in skip_atoms]
bonds = [
b for b in bonds if b.atoms[0] not in skip_atoms
and b.atoms[1] not in skip_atoms
]
residues = struct.residues
# Put the atoms in the order we want for output
if status:
status("Putting atoms in input order")
atoms.sort(key=sort_key_func)
# if anchor is not None, then there will be two entries in
# the @SET section of the file...
if anchor:
sets = 2
else:
sets = 0
# number of entries for various sections...
print("%d %d %d 0 %d" % (len(atoms), len(bonds), len(residues), sets),
file=f)
# type of molecule
if hasattr(struct, "mol2_type"):
mtype = struct.mol2_type
else:
mtype = "SMALL"
from chimerax.atomic import Sequence
for r in struct.residues:
if Sequence.protein3to1(r.name) != 'X':
mtype = "PROTEIN"
break
if Sequence.nucleic3to1(r.name) != 'X':
mtype = "NUCLEIC_ACID"
break
print(mtype, file=f)
# indicate type of charge information
if hasattr(struct, 'charge_model'):
print(struct.charge_model, file=f)
else:
print("NO_CHARGES", file=f)
if hasattr(struct, 'mol2_comment'):
print("\n%s" % struct.mol2_comment, file=f)
else:
print("\n", file=f)
if status:
status("writing atoms")
# atom section header
print("%s" % ATOM_HEADER, file=f)
# make a dictionary of residue indices so that we can do quick look ups
res_indices = {}
for i, r in enumerate(residues):
res_indices[r] = i + 1
for i, atom in enumerate(atoms):
# atom ID, starting from 1
print("%7d" % (i + 1), end=" ", file=f)
# atom name, possibly rearranged if it's a hydrogen
if sybyl_hyd_naming and not atom.name[0].isalpha():
atom_name = atom.name[1:] + atom.name[0]
else:
atom_name = atom.name
print("%-8s" % atom_name, end=" ", file=f)
# use correct relative coordinate position
coord = xform * atom.scene_coord
print("%9.4f %9.4f %9.4f" % tuple(coord), end=" ", file=f)
# atom type
if gaff_type:
try:
atom_type = atom.gaff_type
except AttributeError:
if not gaff_fail_error:
raise
raise gaff_fail_error(
"%s has no Amber/GAFF type assigned.\n"
"Use the AddCharge tool to assign Amber/GAFF types." %
atom)
elif hasattr(atom, 'mol2_type'):
atom_type = atom.mol2_type
elif atom in amide_Ns:
atom_type = "N.am"
elif atom.structure_category == "solvent" \
and atom.residue.name in Residue.water_res_names:
if atom.element.name == "O":
atom_type = "O.t3p"
else:
atom_type = "H.t3p"
elif atom.element.name == "N" and len(
[r for r in atom.rings() if r.aromatic]) > 0:
atom_type = "N.ar"
elif atom.idatm_type == "C2" and len(
[nb for nb in atom.neighbors if nb.idatm_type == "Ng+"]) > 2:
atom_type = "C.cat"
elif sulfur_oxygen(atom):
atom_type = "O.2"
else:
try:
atom_type = chimera_to_sybyl[atom.idatm_type]
except KeyError:
session.logger.warning(
"Atom whose IDATM type has no equivalent"
" Sybyl type: %s (type: %s)" % (atom, atom.idatm_type))
atom_type = str(atom.element)
print("%-5s" % atom_type, end=" ", file=f)
# residue-related info
res = atom.residue
# residue index
print("%5d" % res_indices[res], end=" ", file=f)
# substructure identifier and charge
if hasattr(atom, 'charge') and atom.charge is not None:
charge = atom.charge
else:
charge = 0.0
if substructure_names:
rname = substructure_names[res]
elif res_num:
rname = "%3s%-5d" % (res.name, res.number)
else:
rname = "%3s" % res.name
print("%s %9.4f" % (rname, charge), file=f)
if status:
status("writing bonds")
# bond section header
print("%s" % BOND_HEADER, file=f)
# make an atom-index dictionary to speed lookups
atom_indices = {}
for i, a in enumerate(atoms):
atom_indices[a] = i + 1
for i, bond in enumerate(bonds):
a1, a2 = bond.atoms
# ID
print("%6d" % (i + 1), end=" ", file=f)
# atom IDs
print("%4d %4d" % (atom_indices[a1], atom_indices[a2]),
end=" ",
file=f)
# bond order; give it our best shot...
if hasattr(bond, 'mol2_type'):
print(bond.mol2_type, file=f)
continue
amide_A1 = a1 in amide_CNs
amide_A2 = a2 in amide_CNs
if amide_A1 and amide_A2:
print("am", file=f)
continue
if amide_A1 or amide_A2:
if a1 in amide_Os or a2 in amide_Os:
print("2", file=f)
else:
print("1", file=f)
continue
aromatic = False
# 'bond' might be a metal-coordination bond so do a test for rings
if hasattr(bond, 'rings'):
for ring in bond.rings():
if ring.aromatic:
aromatic = True
break
if aromatic:
print("ar", file=f)
continue
try:
geom1 = idatm_info[a1.idatm_type].geometry
except KeyError:
print("1", file=f)
continue
try:
geom2 = idatm_info[a2.idatm_type].geometry
except KeyError:
print("1", file=f)
continue
# sulfone/sulfoxide is classically depicted as double-
# bonded despite the high dipolar character of the
# bond making it have single-bond character. For
# output, use the classical values.
if sulfur_oxygen(a1) or sulfur_oxygen(a2):
print("2", file=f)
continue
if geom1 not in [2, 3] or geom2 not in [2, 3]:
print("1", file=f)
continue
# if either endpoint atom is in an aromatic ring and
# the bond isn't, it's a single bond...
for endp in [a1, a2]:
aromatic = False
for ring in endp.rings():
if ring.aromatic:
aromatic = True
break
if aromatic:
break
else:
# neither endpoint in aromatic ring
if geom1 == 2 and geom2 == 2:
print("3", file=f)
else:
print("2", file=f)
continue
print("1", file=f)
if status:
status("writing residues")
# residue section header
print("%s" % SUBSTR_HEADER, file=f)
for i, res in enumerate(residues):
# residue id field
print("%6d" % (i + 1), end=" ", file=f)
# residue name field
if substructure_names:
rname = substructure_names[res]
elif res_num:
rname = "%3s%-4d" % (res.name, res.number)
else:
rname = "%3s" % res.name
print(rname, end=" ", file=f)
# ID of the root atom of the residue
chain_atom = res.principal_atom
if chain_atom is None:
# if writing out a selection, not all residue atoms
# might be in atom_indices...
for chain_atom in res.atoms:
if chain_atom in atom_indices:
break
print("%5d" % atom_indices[chain_atom], end=" ", file=f)
print("RESIDUE 4", end=" ", file=f)
# Sybyl seems to use chain 'A' when chain ID is blank,
# so run with that
chain_id = res.chain_id
if not chain_id.strip():
chain_id = 'A'
print("%-4s %3s" % (chain_id, res.name), end=" ", file=f)
# number of out-of-substructure bonds
cross_res_bonds = 0
for a in res.atoms:
for nb in a.neighbors:
if nb.residue != res:
cross_res_bonds += 1
print("%5d" % cross_res_bonds, end="", file=f)
# print "ROOT" if first or only residue of a chain
if not res.chain or res.chain.existing_residues[0] == res:
print(" ROOT", file=f)
else:
print(file=f)
# write flexible ligand docking info
if anchor:
if status:
status("writing anchor info")
print("%s" % SET_HEADER, file=f)
atom_indices = {}
for i, a in enumerate(atoms):
atom_indices[a] = i + 1
bond_indices = {}
for i, b in enumerate(bonds):
bond_indices[b] = i + 1
print(
"ANCHOR STATIC ATOMS <user> **** Anchor Atom Set",
file=f)
print(len(anchor), end=" ", file=f)
for a in anchor:
if a in atom_indices:
print(atom_indices[a], end=" ", file=f)
print(file=f)
print(
"RIGID STATIC BONDS <user> **** Rigid Bond Set",
file=f)
bonds = anchor.intra_bonds
print(len(bonds), end=" ", file=f)
for b in bonds:
if b in bond_indices:
print(bond_indices[b], end=" ", file=f)
print(file=f)
if file_name != f:
f.close()
if status:
status("Wrote Mol2 file %s" % file_name)
def read(session, f):
doing = None
sequences = []
header_ok = False
line_num = 0
align_start_index = None
for line in f.readlines():
if doing == 'alignments':
# don't strip() alignment section since it has significant leading spaces
line = line.rstrip()
else:
line = line.strip()
line_num += 1
if not header_ok:
if line.lower().startswith("hssp"):
header_ok = True
continue
raise FormatSyntaxError("No initial HSSP header line")
if line.startswith('##'):
if doing == 'proteins' and not sequences:
raise FormatSyntaxError("No entries in PROTEINS section")
try:
doing = line.split()[1].lower()
except IndexError:
doing = None
if doing == 'alignments':
try:
hashes, alignments, begin, dash, end = line.strip().split()
begin = int(begin)
end = int(end)
except ValueError:
raise FormatSyntaError("ALIGNMENTS line (line #%d) not of the form: "
"## ALIGNMENTS (number) - (number)" % line_num)
continue
if doing == 'proteins':
if not line[0].isdigit():
continue
try:
seq_name = line.split()[2]
except IndexError:
raise FormatSyntaxError("Line %d in PROTEINS section does not start with "
"[integer] : [sequence name]" % line_num)
sequences.append(Sequence(name=make_readable(seq_name)))
elif doing == 'alignments':
if line.lstrip().lower().startswith('seqno'):
try:
align_start_index = line.index('.')
except Exception:
raise FormatSyntaxError("No indication of alignment starting column "
"('.' character) in SeqNo line in ALIGNMENTS section")
continue
if align_start_index == None:
raise FormatSyntaxError("No initial SeqNo line in ALIGNMENTS section")
block = line[align_start_index:]
if not block:
raise FormatSyntaxError("No alignment block given on line %d" % line_num)
block_len = end - begin + 1
if len(block) > block_len:
raise FormatSyntaxError("Too many characters (%d, only %d sequences) in "
"alignment block given on line %d" % (len(block), block_len, line_num))
block = block + ' ' * (block_len - len(block))
for seq, c in zip(sequences[begin-1:end], block):
seq.append(c)
f.close()
return sequences, {}, {}