def test_makeheader_1(self):
tags = [
'MAINID', '1', {'A', 'B', 'C'}, 'MADEUP',
'A made up protein information'
]
expected_header = '>crops|MAINID_1|Chains A,B,C|Source: MADEUP|A made up protein information'
new_header = cit.makeheader(mainid=tags[0],
seqid=tags[1],
chains=tags[2],
source=tags[3],
extrainfo=tags[4])
self.assertEqual(new_header, expected_header)
def update_cropsheader(self):
"""Update `cropsheader`. Useful after updating any information from the sequence."""
if self.oligomer_id is None:
tag1 = 'NoID'
else:
tag1 = self.oligomer_id
tag2 = self.infostring
if self.ncrops() == 0 and 'cropseq' not in self.seqs:
pass
else:
if tag2[-1] != "|":
tag2 += '|'
tag2 += self.cropinfo()
self.crops_header = makeheader(mainid=tag1,
seqid=self.name,
chains=self.chains,
source=self.source,
extrainfo=tag2)
def __repr__(self):
chtype = self.biotype if self.biotype is not None else 'Undefined'
if 'mainseq' not in self.seqs:
logging.critical("'mainseq' sequence not found.")
raise ValueError
if len(self.seqs['mainseq']) <= 20:
showseq = self.seqs['mainseq']
else:
showseq = (self.seqs['mainseq'][:10] + '[...]' +
self.seqs['mainseq'][len(self.seqs['mainseq']) - 10:])
tempolig = self.oligomer_id if self.oligomer_id is not None else 'NOID'
shortid = makeheader(mainid=tempolig,
seqid=self.name,
chains=self.chains,
short=True)
string = (self._kind + " object " + shortid + " (seq=" + str(showseq) +
", type=" + chtype + ", length=" +
str(len(self.seqs['mainseq'])) + ")")
return string
def dumpmap(self, out, split=False):
"""Write header and cropmap to a file. If file exists, output is appended.
:param out: An output filepath (str) or an open file.
:type out: str, file
:param backmap: If True, the output will be self.cropbackmap, defaults to False.
:type backmap: bool, optional
:param split: If True, identical maps are dumped for every chain, defaults to False.
:type split: bool, optional
:raises `TypeError`: If `out` is neither a string nor an open file.
:raises `ValueError`: If one or both of `cropmap` and `cropbackmap` are empty.
:raises `KeyError`: If object contains no chains.
"""
if not isinstance(out, str) and not isinstance(out, io.IOBase):
logging.critical("Argument 'out' should be a string or a file.")
raise TypeError
if self.cropmap is None:
stringerr = "Cropmap not found in sequence."
logging.critical(stringerr)
raise ValueError
if (self.chains is None
or (isinstance(self.chains, set) and len(self.chains) == 0)):
logging.critical('No chains defined in sequence.')
raise KeyError
outheader = []
if split:
chset = []
for ch in self.chains:
chset.append({ch})
else:
chset = [self.chains]
if self.oligomer_id is None:
tag1 = 'NoID'
else:
tag1 = self.oligomer_id
tag2 = self.infostring
if self.ncrops() == 0 and 'cropseq' not in self.seqs:
pass
else:
if self.infostring[-1] != "|":
tag2 += '|'
tag2 += self.cropinfo()
for ch in chset:
outheader.append(
makeheader(mainid=tag1,
seqid=self.name,
chains=ch,
source=self.source,
extrainfo=tag2))
output = ''
for header in outheader:
if isinstance(out, io.IOBase):
out.write(header + '\n')
for key, value in self.cropmap.items():
if value is not None:
out.write(str(key) + ' ' + str(value) + '\n')
else:
out.write(str(key) + ' 0\n')
else:
output += header + os.linesep
for key, value in self.cropmap.items():
if value is not None:
output += str(key) + ' ' + str(value) + os.linesep
else:
output += str(key) + ' 0' + os.linesep
if isinstance(out, io.IOBase) is False:
if out.lower() == 'string':
return output
else:
outpath = out
op = 'a' if os.path.isfile(outpath) else 'w'
with open(outpath, op) as out:
out.write(output)
return
def dump(self, out, split=False, oneline=False):
"""Write header and main sequence to a file. If the file exists, output is appended.
:param out: An output filepath (str), 'string', or an open file.
:type out: str, file
:param split: If True, identical sequences are dumped for every chain, defaults to False.
:type split: bool, optional
:param oneline: If True, sequences are not split in 80 residue-lines, defaults to False.
:type oneline: bool, optional
:raises `TypeError`: If `out` is neither a string nor an open file.
:raises `KeyError`: If object contains no chains.
:return: A string containing the output if and only if out=='string'.
:rtype: str
"""
if not isinstance(out, str) and not isinstance(out, io.IOBase):
logging.critical("Argument 'out' should be a string or a file.")
raise TypeError
if (self.chains is None
or (isinstance(self.chains, set) and len(self.chains) == 0)):
logging.critical('No chains defined in sequence.')
raise KeyError
outheader = []
if split:
chset = []
for ch in self.chains:
chset.append({ch})
else:
chset = [self.chains]
if self.oligomer_id is None:
tag1 = 'NoID'
else:
tag1 = self.oligomer_id
tag2 = self.infostring
if self.ncrops() == 0 and 'cropseq' not in self.seqs:
pass
else:
if self.infostring[-1] != "|":
tag2 += '|'
tag2 += self.cropinfo()
for ch in chset:
outheader.append(
makeheader(mainid=tag1,
seqid=self.name,
chains=ch,
source=self.source,
extrainfo=tag2))
if not oneline:
lenseq = len(self.seqs['mainseq'])
nlines = int((lenseq - 1) / 80) + 1
output = ''
for header in outheader:
if isinstance(out, io.IOBase) is True:
out.write(header + '\n')
if oneline:
out.write(self.seqs['mainseq'] + '\n')
else:
for n in range(nlines):
out.write(self.seqs['mainseq'][n * 80:(n + 1) * 80] +
'\n')
else:
output += header + os.linesep
if oneline:
output += self.seqs['mainseq'] + os.linesep
else:
for n in range(nlines):
output += self.seqs['mainseq'][n * 80:(n + 1) *
80] + os.linesep
if isinstance(out, io.IOBase) is False:
if out.lower() == 'string':
return output
else:
outpath = out
op = 'a' if os.path.isfile(outpath) else 'w'
with open(outpath, op) as out:
out.write(output)
return
def __init__(self,
seqid=None,
oligomer=None,
seq=None,
chains=None,
source=None,
header=None,
biotype=None,
extrainfo=None):
self.oligomer_id = None
self.name = None
self.chains = set()
self.source = None
self.source_headers = []
self.crops_header = None
self.seqs = {}
self.biotype = None
self.infostring = None
self.cropmap = None
self.cropbackmap = None
self.msa = None
self.cropmsa = None
self.intervals = None
if header is not None:
if isinstance(header, str):
self.source_headers.append(header)
try:
header_info = retrieve_id(header)
except Exception:
logging.warning(
'Header format not recognised. Information not extracted.'
)
header_info = None
else:
logging.critical("Argument 'header' should be a string.")
raise TypeError
else:
header_info = None
if seqid is not None:
if isinstance(seqid, str):
self.name = seqid
elif isinstance(seqid, int):
self.name = str(seqid)
else:
logging.critical("Sequence ID 'seqid' should be a string.")
raise TypeError
else:
if header_info is not None:
if 'seqid' in header_info:
self.name = header_info['seqid']
else:
self.name = '1'
if seq is not None:
if isinstance(seq, str):
self.seqs['mainseq'] = seq
else:
logging.critical("Chain sequence 'seq' should be a string.")
raise TypeError
else:
self.seqs['mainseq'] = ''
if oligomer is not None:
if isinstance(oligomer, str):
self.oligomer_id = oligomer
else:
logging.critical("Oligomer ID 'oligomer' should be a string.")
raise TypeError
else:
if header_info is not None:
if 'mainid' in header_info:
self.oligomer_id = header_info['mainid']
if chains is not None:
if isinstance(chains, set):
for ch in chains:
if isinstance(ch, str):
self.chains.add(ch)
else:
logging.critical(
"Chain IDs in 'chains' set should be strings.")
raise TypeError
else:
logging.critical(
"Argument 'chains' should be a set of strings.")
raise TypeError
else:
if header_info is not None:
if 'chains' in header_info:
self.chains = header_info['chains']
if source is not None:
if isinstance(source, str):
self.source = source
else:
logging.critical("Argument 'source' should be a string.")
raise TypeError
else:
if header_info is not None:
if 'source' in header_info:
self.source = header_info['source']
if biotype is not None:
if biotype.lower() == 'guess':
self.biotype = guess_type(seq)
else:
self.biotype = biotype
else:
self.biotype = None
if extrainfo is not None:
if isinstance(extrainfo, str):
self.infostring = extrainfo
else:
logging.critical("Argument 'extrainfo' should be a string.")
raise TypeError
else:
if header_info is not None:
if 'comments' in header_info:
self.infostring = header_info['comments']
else:
self.infostring = ""
if oligomer is None:
self.crops_header = makeheader(mainid='NOID',
seqid=self.name,
chains=self.chains,
source=self.source,
extrainfo=self.infostring)
else:
self.crops_header = makeheader(mainid=self.oligomer_id,
seqid=self.name,
chains=self.chains,
source=self.source,
extrainfo=self.infostring)