def buildMSA(sequences, title='Unknown', labels=None, **kwargs):
"""
Aligns sequences with clustalw or clustalw2 and returns the resulting MSA.
:arg sequences: a file, MSA object or a list or array containing sequences
as Atomic objects with :func:`getSequence` or Sequence objects or strings.
If strings are used then labels must be provided using ``labels``
:type sequences: :class:`Atomic`, :class:`.MSA`,
:class:`~numpy.ndarray`, str
:arg title: the title for the MSA and it will be used as the prefix for output files.
:type title: str
:arg labels: a list of labels to go with the sequences
:type labels: list
:arg align: whether to align the sequences
default True
:type align: bool
:arg method: alignment method, one of either biopython.align.globalms or clustalw(2).
default 'clustalw'
:type align: str
"""
align = kwargs.get('align', True)
method = kwargs.pop('method', 'clustalw')
# 1. check if sequences are in a fasta file and if not make one
if isinstance(sequences, str):
filename = sequences
elif not isinstance(sequences, MSA):
try:
max_len = 0
for sequence in sequences:
if isinstance(sequence, Atomic):
if len(sequence.ca.copy()) > max_len:
max_len = len(sequence.ca.copy())
elif isinstance(sequence, MSA):
if len(sequence[0]) > max_len:
max_len = len(sequence[0])
else:
if len(sequence) > max_len:
max_len = len(sequence)
msa = []
fetched_labels = []
for i, sequence in enumerate(sequences):
if isinstance(sequence, Atomic):
strseq = sequence.ca.getSequence()
label = sequence.getTitle()
elif isinstance(sequence, Sequence):
strseq = str(sequence)
label = sequence.getLabel()
elif isinstance(sequence, MSA):
strseq = str(sequence[0])
label = sequence.getLabel(0)
LOGGER.warn(
'Only the first sequence in the MSA at entry {0} is used.'
.format(i))
elif isinstance(sequence, str):
strseq = sequence
label = str(i + 1)
else:
raise TypeError('sequences should be a list of strings, '
'Atomic, or Sequence instances')
strseq = strseq + '-' * (max_len - len(strseq))
msa.append(array(list(strseq)))
fetched_labels.append(label)
sequences = array(msa)
except:
raise TypeError('sequences should be iterable')
# "if a list" is a pythonic way to check if a list is empty or not (or none)
if not labels and fetched_labels:
labels = fetched_labels
label = [label.replace(' ', '_') for label in labels]
# labels checkers are removed because they will be properly handled in MSA class initialization
msa = MSA(msa=sequences, title=title, labels=labels)
if align and 'clustal' in method:
filename = writeMSA(title + '.fasta', msa)
if align:
# 2. find and run alignment method
if 'biopython' in method:
if len(sequences) == 2:
msa, _, _ = alignTwoSequencesWithBiopython(
sequences[0], sequences[1], **kwargs)
else:
raise ValueError(
"Provide only two sequences or another method. \
Biopython pairwise alignment can only be used \
to build an MSA with two sequences.")
elif 'clustalw' in method:
clustalw = which('clustalw')
if clustalw is None:
if which('clustalw2') is not None:
clustalw = which('clustalw2')
else:
raise EnvironmentError(
"The executable for clustalw was not found, \
install clustalw or add it to the path."
)
os.system('"%s" %s -OUTORDER=INPUT' % (clustalw, filename))
# 3. parse and return the new MSA
msa = parseMSA(title + '.aln')
else:
alignTool = which(method)
if alignTool is None:
raise EnvironmentError("The executable for {0} was not found, \
install it or add it to the path.".
format(alignTool))
os.system('"%s" %s -OUTORDER=INPUT' % (clustalw, filename))
# 3. parse and return the new MSA
msa = parseMSA(title + '.aln')
return msa
def alignSequencesByChain(PDBs, **kwargs):
"""
Runs buildMSA for each chain and optionally joins the results.
Returns either a single MSA or a dictionary containing an MSA for each chain.
:arg PDBs: a list or array of :class:`AtomGroup` objects or PDB IDs
a mixed list containing both is acceptable
:type PDBs: list or :class:`~numpy.ndarray`
:arg join_chains: whether to join chain alignments
default is True
:type join_chains: bool
:arg join_char: a character for joining chain alignments
default is '/' as used by PIR format alignments
:type join_char: str
"""
if not (isinstance(PDBs, list) or isinstance(PDBs, ndarray)):
raise TypeError('PDBs should be a list or array')
if PDBs == []:
raise ValueError('PDBs should not be an empty list')
pdbs = []
chains = []
for i, pdb in enumerate(PDBs):
if isinstance(pdb, Atomic):
pdbs.append(pdb)
else:
raise TypeError(
'each entry in PDBs must be a :class:`Atomic` instance')
chains.append([])
for chain in list(pdbs[i].getHierView()):
chains[i].append(chain)
if i != 0 and len(chains[i]) != len(chains[0]):
raise ValueError('all pdbs should have the same number of chains')
labels = []
for pdb in pdbs:
chids = ''
for chain in list(pdb.getHierView()):
chids += chain.getChid()
labels.append(pdb.getTitle().split('_')[0] + '_' + chids)
chains = array(chains)
chain_alignments = []
alignments = {}
labels_lists = []
for j in range(len(chains[0])):
prefix = 'chain_' + chains[0, j].getChid()
msa = buildMSA(chains[:, j], title=prefix, labels=labels)
# make all alignments have the sequences in the same order as the 0th
labels_lists.append([])
for sequence in msa:
labels_lists[j].append(sequence.getLabel())
if j > 0:
msaarr = []
for label in labels_lists[0]:
msaarr.append(msa.getArray()[msa.getIndex(label)])
msaarr = array(msaarr)
msa = MSA(msaarr,
title='reordered_msa_1',
labels=list(labels_lists[0]))
writeMSA(prefix + '.aln', msa)
chain_alignments.append(msa)
# after reordering, create the alignments dictionary
alignments[labels_lists[0][0].split('_')[1][j]] = msa
join_chains = kwargs.get('join_chains', True)
join_char = kwargs.get('join_char', '/')
if join_chains:
aligned_sequences = list(zeros(shape(chain_alignments)).T)
for j in range(shape(chain_alignments)[1]):
aligned_sequences[j] = list(aligned_sequences[j])
orig_labels = []
for i, chain_alignment in enumerate(chain_alignments):
for j, sequence in enumerate(chain_alignment):
aligned_sequences[j][i] = str(sequence)
if i == 0:
orig_labels.append(sequence.getLabel())
joined_msaarr = []
for j in range(shape(chain_alignments)[1]):
joined_msaarr.append(
array(list(join_char.join(aligned_sequences[j]))))
joined_msaarr = array(joined_msaarr)
result = MSA(joined_msaarr, title='joined_chains', labels=orig_labels)
result = refineMSA(result, colocc=1e-9) # remove gap-only cols
else:
result = alignments
return result
def buildMSA(sequences, title='Unknown', labels=None, **kwargs):
"""
Aligns sequences with clustalw or clustalw2 and returns the resulting MSA.
:arg sequences: a file, MSA object or a list or array containing sequences
as Atomic objects with :func:`getSequence` or Sequence objects or strings.
If strings are used then labels must be provided using ``labels``
:type sequences: :class:`Atomic`, :class:`.MSA`,
:class:`~numpy.ndarray`, str
:arg title: the title for the MSA and it will be used as the prefix for output files.
:type title: str
:arg labels: a list of labels to go with the sequences
:type labels: list
:arg align: whether to do alignment with clustalw(2)
default True
:type align: bool
"""
align = kwargs.get('align', True)
# 1. check if sequences are in a fasta file and if not make one
if isinstance(sequences, str):
filename = sequences
elif not isinstance(sequences, MSA):
try:
max_len = 0
for sequence in sequences:
if len(sequence) > max_len:
max_len = len(sequence)
msa = []
fetched_labels = []
for i, sequence in enumerate(sequences):
if isinstance(sequence, Atomic):
strseq = sequence.getSequence()
label = sequence.getTitle()
elif isinstance(sequence, Sequence):
strseq = str(sequence)
label = sequence.getLabel()
elif isinstance(sequence, str):
strseq = sequence
label = str(i + 1)
else:
raise TypeError('sequences should be a list of strings, '
'Atomic, or Sequence instances')
strseq = strseq + '-' * (max_len - len(strseq))
msa.append(array(list(strseq)))
fetched_labels.append(label)
sequences = array(msa)
except:
raise TypeError('sequences should be iterable')
# "if a list" is a pythonic way to check if a list is empty or not (or none)
if not labels and fetched_labels:
labels = fetched_labels
# labels checkers are removed because they will be properly handled in MSA class initialization
msa = MSA(msa=sequences, title=title, labels=labels)
if align:
filename = writeMSA(title + '.fasta', msa)
if align:
# 2. find and run alignment method
clustalw = which('clustalw')
if clustalw is None:
if which('clustalw2') is not None:
clustalw = which('clustalw2')
else:
raise EnvironmentError(
"The executable for clustalw was not found, \
install clustalw or add it to the path."
)
os.system('"%s" %s' % (clustalw, filename))
# 3. parse and return the new MSA
msa = parseMSA(title + '.aln')
return msa
def buildMSA(sequences, title='Unknown', labels=None, **kwargs):
"""
Aligns sequences with clustalw or clustalw2 and returns the resulting MSA.
:arg sequences: a file, MSA object or a list or array containing sequences
as Atomic objects with :func:`getSequence` or Sequence objects or strings.
If strings are used then labels must be provided using ``labels``
:type sequences: :class:`Atomic`, :class:`.MSA`,
:class:`~numpy.ndarray`, str
:arg title: the title for the MSA and it will be used as the prefix for output files.
:type title: str
:arg labels: a list of labels to go with the sequences
:type labels: list
:arg align: whether to align the sequences
default True
:type align: bool
:arg method: alignment method, one of either biopython.align.globalms or clustalw(2).
default 'clustalw'
:type align: str
"""
align = kwargs.get('align', True)
method = kwargs.pop('method', 'clustalw')
# 1. check if sequences are in a fasta file and if not make one
if isinstance(sequences, str):
filename = sequences
elif not isinstance(sequences, MSA):
try:
max_len = 0
for sequence in sequences:
if isinstance(sequence, Atomic):
if len(sequence.ca.copy()) > max_len:
max_len = len(sequence.ca.copy())
elif isinstance(sequence, MSA):
if len(sequence[0]) > max_len:
max_len = len(sequence[0])
else:
if len(sequence) > max_len:
max_len = len(sequence)
msa = []
fetched_labels = []
for i, sequence in enumerate(sequences):
if isinstance(sequence, Atomic):
strseq = sequence.ca.getSequence()
label = sequence.getTitle()
elif isinstance(sequence, Sequence):
strseq = str(sequence)
label = sequence.getLabel()
elif isinstance(sequence, MSA):
strseq = str(sequence[0])
label = sequence.getLabel(0)
LOGGER.warn('Only the first sequence in the MSA at entry {0} is used.'
.format(i))
elif isinstance(sequence, str):
strseq = sequence
label = str(i + 1)
else:
raise TypeError('sequences should be a list of strings, '
'Atomic, or Sequence instances')
strseq = strseq + '-'*(max_len - len(strseq))
msa.append(array(list(strseq)))
fetched_labels.append(label)
sequences = array(msa)
except:
raise TypeError('sequences should be iterable')
# "if a list" is a pythonic way to check if a list is empty or not (or none)
if not labels and fetched_labels:
labels = fetched_labels
label = [label.replace(' ','_') for label in labels]
# labels checkers are removed because they will be properly handled in MSA class initialization
msa = MSA(msa=sequences, title=title, labels=labels)
if align and 'clustal' in method:
filename = writeMSA(title + '.fasta', msa)
if align:
# 2. find and run alignment method
if 'biopython' in method:
if len(sequences) == 2:
msa, _, _ = alignTwoSequencesWithBiopython(sequences[0], sequences[1], **kwargs)
else:
raise ValueError("Provide only two sequences or another method. \
Biopython pairwise alignment can only be used \
to build an MSA with two sequences.")
elif 'clustalw' in method:
clustalw = which('clustalw')
if clustalw is None:
if which('clustalw2') is not None:
clustalw = which('clustalw2')
else:
raise EnvironmentError("The executable for clustalw was not found, \
install clustalw or add it to the path.")
os.system('"%s" %s -OUTORDER=INPUT'%(clustalw, filename))
# 3. parse and return the new MSA
msa = parseMSA(title + '.aln')
else:
alignTool = which(method)
if alignTool is None:
raise EnvironmentError("The executable for {0} was not found, \
install it or add it to the path.".format(alignTool))
os.system('"%s" %s -OUTORDER=INPUT'%(clustalw, filename))
# 3. parse and return the new MSA
msa = parseMSA(title + '.aln')
return msa