class Test(BT.BiskitTest):
"""DSSP test"""
TAGS = [BT.EXE]
def prepare(self):
self.f = T.testRoot()+"/com/1BGS.pdb"
def test_DSSP( self ):
"""DSSP test"""
from Biskit import PDBModel
if self.local: print 'Loading PDB...'
self.m = PDBModel(self.f)
self.m = self.m.compress( self.m.maskProtein() )
if self.local: print 'Starting DSSP'
self.dssp = Dssp( self.m )
if self.local: print 'Running DSSP'
self.result = self.dssp.run()
if self.local:
print "Sequence :", self.m.sequence()
print "Secondary:", self.result
self.assertEquals( self.result, self.EXPECTED)
EXPECTED = '.....SHHHHHHHHHHHSS..TTEE.HHHHHHHT..GGGT.HHHHSTT.EEEEEEE..TT..S...TT..EEEEE.S..SSS..S.EEEEETT..EEEESSSSSS.EE...EEEEETTT..SHHHHHHHHHHHHT..TT..SSHHHHHHHHHHT..SSEEEEEE.HHHHHHHTTTTHHHHHHHHHHHHHHT..EEEEE.'
class Test(BT.BiskitTest):
"""DSSP test"""
TAGS = [BT.EXE]
def prepare(self):
self.f = T.testRoot() + "/com/1BGS.pdb"
def test_DSSP(self):
"""DSSP test"""
from Biskit import PDBModel
if self.local: print 'Loading PDB...'
self.m = PDBModel(self.f)
self.m = self.m.compress(self.m.maskProtein())
if self.local: print 'Starting DSSP'
self.dssp = Dssp(self.m, verbose=self.local, debug=self.DEBUG)
if self.local: print 'Running DSSP'
self.result = self.dssp.run()
if self.local:
print "Sequence :", self.m.sequence()
print "Secondary:", self.result
self.assertEquals(self.result, self.EXPECTED)
EXPECTED = '.....SHHHHHHHHHHHSS..TTEE.HHHHHHHT..GGGT.HHHHSTT.EEEEEEE..TT..S...TT..EEEEE.S..SSS..S.EEEEETT..EEEESSSSSS.EE...EEEEETTT..SHHHHHHHHHHHHT..TT..SSHHHHHHHHHHT..SSEEEEEE.HHHHHHHTTTTHHHHHHHHHHHHHHT..EEEEE.'
class Test(BT.BiskitTest):
"""DSSP test"""
TAGS = [BT.EXE]
def prepare(self):
self.f = T.testRoot() + "/com/1BGS.pdb"
def generic_test(self, pdb, expected=None, proteinonly=False):
"""generic DSSP test"""
from Biskit import PDBModel
if self.local: print 'Loading PDB...'
self.m = PDBModel(pdb)
if proteinonly:
self.m = self.m.compress(self.m.maskProtein())
if self.local: print 'Starting DSSP'
self.dssp = Dssp(self.m, verbose=self.local, debug=self.DEBUG)
if self.local: print 'Running DSSP'
self.result = self.dssp.run()
if self.local:
print "Sequence :", self.m.sequence()
print "Secondary:", self.result
if expected:
self.assertEquals(self.result, expected)
EXPECTED = '.....SHHHHHHHHHHHSS..TTEE.HHHHHHHT..GGGT.HHHHSTT.EEEEEEE..TT..S...TT..EEEEE.S..SSS..S.EEEEETT..EEEESSSSSS.EE...EEEEETTT..SHHHHHHHHHHHHT..TT..SSHHHHHHHHHHT..SSEEEEEE.HHHHHHHTTTTHHHHHHHHHHHHHHT..EEEEE.'
def test_DSSP(self):
"""DSSP test"""
self.generic_test(self.f, expected=self.EXPECTED, proteinonly=True)
def test_DSSP_2W3A(self):
"""DSSP test 2W3A"""
self.generic_test('2W3A', proteinonly=True)
def test_DSSP_1R4Q(self):
"""DSSP test"""
self.generic_test('1R4Q', proteinonly=True)
def test_DSSP_1AM7(self):
self.generic_test('1AM7')
def test_DSSP_5FB8(self):
self.generic_test('5fb8')
def process_all( self, file_dic, keep_hetatoms=0 ):
"""
Process PDB files in file_dic.
The PDB is read from:
- L{TemplateSearcher.F_NR}
and cleaned files are written to:
- L{F_CLEANED}
- L{F_COFFEE}
- L{F_MODELLER}
If the file L{F_CLEANED} already exists, this file is
used to write modeller and t-coffee pdbs.
@param file_dic: dictionary mapping filenames of pdb files to
the chains of interest, e.g. { fname : chain_id, }
@type file_dic: {str:str}
@param keep_hetatoms: keep hetatoms (default: 0)
@type keep_hetatoms: 0|1
"""
fasta = ""
c = None
for f, id in file_dic.items():
self.logWrite( '\nCLEANING ' + f + '...')
try:
code = string.split(f, '/')[-1][:4]
## clean to standard PDB
c = PDBCleaner( f, self.log )
## if /templates/nr_cleaned/pfb-file exists
if os.path.exists(
self.outFolder+self.F_CLEANED + code + '.pdb' ):
model = PDBModel( self.outFolder+self.F_CLEANED \
+ code + '.pdb' )
## read pdb from /templates/nr
else:
model = c.process( keep_hetatoms=keep_hetatoms )
## write complete cleaned PDB
model.writePdb( self.outFolder+self.F_CLEANED \
+ code + '.pdb')
code = model.pdbCode
title = code + '_' + id
## extract needed chain
if len( id ) > 0:
model = self.__chain_by_id( model, id )
fname = "%s%s.pdb" % (self.outFolder + self.F_MODELLER, title)
self.write_modeller_pdb( model, fname )
## write T_coffee CA trace with res sequence
fname = "%s%s.alpha"%(self.outFolder + self.F_COFFEE, title)
self.write_tcoffee_pdb( model, fname )
## append fasta
fasta += self.fasta_sequence( title, model.sequence() )
except:
self.logWrite( 'Error cleaning ' + f)
self.logWrite( T.lastError() )
self.err_cleaner = c
self.err_file = f
fasta_out = open( self.outFolder + self.F_FASTA, 'w' )
fasta_out.write( fasta )
fasta_out.close()
Syntax pdb2seq.py |pdb_file|
"""
def printSequence( header, s ):
n_chunks = len( s ) / 80
print ">%s" % header
for i in range(0, n_chunks+1):
if i * 80 + 80 < len( s ):
chunk = s[i * 80 : i * 80 + 80]
else:
chunk = s[i * 80 :]
print chunk
if __name__ == '__main__':
if len( sys.argv ) < 1:
_use()
m = PDBModel( sys.argv[1] )
seq = m.sequence()
printSequence( m.pdbCode, seq )
class Test(BT.BiskitTest):
"""DSSP test"""
TAGS = [BT.EXE]
def prepare(self):
self.f = T.testRoot()+"/com/1BGS.pdb"
def generic_test(self, pdb, expected=None, proteinonly=False):
"""generic DSSP test"""
from Biskit import PDBModel
if self.local: print 'Loading PDB...'
self.m = PDBModel(pdb)
if proteinonly:
self.m = self.m.compress( self.m.maskProtein() )
if self.local: print 'Starting DSSP'
self.dssp = Dssp( self.m, verbose=self.local, debug=self.DEBUG )
if self.local: print 'Running DSSP'
self.result = self.dssp.run() ## returns modified PDBModel
self.result = self.result.compress( self.result.maskProtein() )
self.result = ''.join(self.result['dssp'])
if self.local:
print "Sequence :", self.m.sequence()
print "Secondary:", self.result
if expected:
self.assertEquals( self.result, expected)
EXPECTED = '.....SHHHHHHHHHHHSS..TTEE.HHHHHHHT..GGGT.HHHHSTT.EEEEEEE..TT..S...TT..EEEEE.S..SSS..S.EEEEETT..EEEESSSSSS.EE...EEEEETTT..SHHHHHHHHHHHHT..TT..SSHHHHHHHHHHT..SSEEEEEE.HHHHHHHTTTTHHHHHHHHHHHHHHT..EEEEE.'
def test_DSSP( self ):
"""DSSP test"""
self.generic_test(self.f, expected=self.EXPECTED, proteinonly=False)
def test_DSSP_2W3A( self ):
"""DSSP test 2W3A"""
self.generic_test('2W3A', proteinonly=False)
def test_DSSP_3TGI( self ):
"""DSSP test 3TGI"""
## handling of residues with insertion code (184A, 186A, 221A)
self.generic_test('3TGI', proteinonly=False)
def test_DSSP_1R4Q( self ):
"""DSSP test"""
self.generic_test('1R4Q', proteinonly=False)
def test_DSSP_1AM7(self):
self.generic_test('1AM7')
def test_DSSP_5FB8(self):
self.generic_test('5fb8')
def test_DSSP_3j4q(self):
self.generic_test(T.testRoot('dssp') + '/3j4q.pdb')
Syntax pdb2seq.py |pdb_file|
"""
def printSequence( header, s ):
n_chunks = len( s ) / 80
print ">%s" % header
for i in range(0, n_chunks+1):
if i * 80 + 80 < len( s ):
chunk = s[i * 80 : i * 80 + 80]
else:
chunk = s[i * 80 :]
print chunk
if __name__ == '__main__':
if len( sys.argv ) < 1:
_use()
m = PDBModel( sys.argv[1] )
seq = m.sequence()
printSequence( m.pdbCode, seq )