def from_string(cls, domainlist):
"""
Creates a new entry from a CathDomainList string
"""
domainlist = domainlist.strip()
cols = domainlist.split()
if len(cols) != 12:
raise err.ParseError(
"expected 12 cols in CathDomainList line '{}' (found {})".
format(domainlist, len(cols)))
return cls(
domain_id=str(cols[0]),
class_code=int(cols[1]),
arch_code=int(cols[2]),
top_code=int(cols[3]),
homol_code=int(cols[4]),
s35_code=int(cols[5]),
s60_code=int(cols[6]),
s95_code=int(cols[7]),
s100_code=int(cols[8]),
domain_code=int(cols[9]),
atom_length=int(cols[10]),
resolution=float(cols[11]),
)
def from_string(cls, nameline):
nameline = nameline.strip()
cols = nameline.split(sep=None, maxsplit=2)
if len(cols) != 3:
raise err.ParseError(
"expected 3 cols in CathNames line '{}' (found {})".format(
nameline, len(cols)))
name = cols[2]
if not name.startswith(':'):
raise err.ParseError(
"expected name '{}' to start with ':'".format(nameline))
name = name[1:]
return cls(cath_id=str(cols[0]),
example_domain_id=str(cols[1]),
name=str(name))
def run_fasta(self, fasta_file):
"""
Returns scorecons data for the provided FASTA file.
Returns:
ScoreconsResult: scorecons result
"""
tmp_scorecons_file = tempfile.NamedTemporaryFile(mode='w+',
suffix='.scorecons',
delete=True)
scorecons_args = (self.scorecons_path, '-a', fasta_file, '-m',
self.matrix_path, '-o', tmp_scorecons_file.name)
LOG.debug("running scorecons: sys: " + " ".join(scorecons_args))
try:
p = Popen(scorecons_args,
stdout=PIPE,
stderr=PIPE,
universal_newlines=True)
scorecons_out, scorecons_err = p.communicate()
except subprocess.CalledProcessError as e:
LOG.error(
'CMD: %s\nCODE: %s\nOUTPUT: %s\nSTDERR: "%s"\nSTDOUT: "%s"\n',
e.cmd, e.returncode, e.output, e.stderr, e.stdout)
raise e
except:
raise FileNotFoundError(
"Encountered error running scorecons: `{}`".format(
" ".join(scorecons_args)))
match = re.search(r'^DOPS score:\s+([0-9.]+)',
scorecons_out,
flags=re.MULTILINE)
if not match:
raise err.ParseError(
'Failed to find DOPS score in scorecons output: {} (STDERR: {})'
.format(scorecons_out, scorecons_err))
dops_score = float(match.group(1))
sc_numbers = self.__class__.split_scorecons_file(
tmp_scorecons_file.name)
res = ScoreconsResult(dops=dops_score, scores=sc_numbers)
return res
def from_string(cls, domall_line):
"""
Create a new instance from a Domall string
Usage:
::
domall_str = '10gsA D02 F01 2 A 2 - A 78 - A 187 - A 208 - 1 A 79 - A 186 - A 209 - A 209 - (1)'
domall = Domall.from_string(domall_str)
domall.domains[0].segments[0].chain_code # 'A'
domall.domains[0].segments[0].start_pdb # 2
domall.domains[0].segments[0].start_insert # None
domall.domains[0].segments[0].end_pdb # 78
domall.domains[0].segments[0].end_insert # None
"""
domall_line = domall_line.strip()
cols = domall_line.split()
chain_id, dom_count, frag_count = cols[0:3]
dom_count = int(dom_count[1:])
frag_count = int(frag_count[1:])
domains = []
fragments = []
idx = 3
dom_idx = 0
while dom_idx < dom_count:
seg_count = int(cols[idx])
idx += 1
segments = []
# LOG.info("dom[%s] seg_count=%s", dom_idx, seg_count)
for seg_idx in range(seg_count):
start_chain, start_pdb, start_ins, end_chain, end_pdb, end_ins = cols[idx:idx+6]
if start_ins == '-':
start_ins = None
if end_ins == '-':
end_ins = None
idx += 6
seg = Segment(chain_code=start_chain, start_pdb=start_pdb, start_insert=start_ins,
end_pdb=end_pdb, end_insert=end_ins)
# LOG.info("seg[%s]: %s", seg_idx, seg.__dict__)
segments.extend([seg])
dom = Domain(segments=segments)
domains.extend([dom])
dom_idx += 1
frag_idx = 0
while frag_idx < frag_count:
start_chain, start_pdb, start_ins, end_chain, end_pdb, end_ins, frag_len = cols[
idx:idx+7]
idx += 7
frag_idx += 1
frag_len_match = re.match(r'^\((\d+)\)', frag_len)
if not frag_len_match:
raise err.ParseError(
'failed to parse frag len from "{}": {} (idx={})'.format(frag_len, domall_line, idx))
atom_length = frag_len_match.group(1)
frag = Fragment(chain_code=start_chain, start_pdb=start_pdb, start_insert=start_ins,
end_pdb=end_pdb, end_insert=end_ins, atom_length=atom_length)
fragments.extend([frag])
if idx != len(cols):
raise err.ParseError('col index is {}, but there are {} columns: {}'.format(
idx, len(cols), domall_line,
))
return cls(
chain_id=str(chain_id),
domains=domains,
fragments=fragments,
)