def CheckLink(self, query_nid, query_from, query_to, sbjct_nid, sbjct_from,
sbjct_to):
"""check, whether link is faithfull.
"""
query_profile = self.GetProfile(query_nid)
query_profile.useSegment(query_from, query_to)
sbjct_profile = self.GetProfile(sbjct_nid)
sbjct_profile.useSegment(sbjct_from, sbjct_to)
alignator = alignlib.makeFullDP(self.mGop, self.mGep)
result = alignlib.makeAlignataVector()
alignator.Align(query_profile, sbjct_profile, result)
if self.mLogLevel >= 3:
print "# --> %i vs %i: score=%5.2f, length=%i, numgaps=%i, row_from=%i, row_to=%i, col_from=%i, col_to=%i" %\
(query_nid, sbjct_nid,
result.getScore(),
result.getLength(),
result.getNumGaps(),
result.getRowFrom(), result.getRowTo(),
result.getColFrom(), result.getColTo())
sys.stdout.flush()
query_profile.useFullLength()
sbjct_profile.useFullLength()
if result.getScore() > self.mMinAlignmentScore:
return 1, result
else:
return 0, result
fragment.getColFrom(), fragment.getColTo(),
overlap, xcol_from, xcol_to,
(xcol_to - xcol_from) - (col_to - col_from)) )
return ("0",) * 12
##--------------------------------------------------------------------------------------
if __name__ == '__main__':
dbhandle = Pairsdb()
if not dbhandle.Connect():
print "Connection failed"
sys.exit(1)
a = alignlib.makeFullDP( -10.0, -2.0 )
alignator = Alignator( a )
x = ScopTesterSequences( dbhandle, alignator )
x.Process()
if param_alignator == 0:
a = alignlib.makeFullDP( param_gop, param_gep)
alignator = Alignator( a )
if param_entities == 0:
tester = ScopTesterSequences( dbhandle, alignator )
tester.mLogLevel = param_loglevel
def CheckLink(self, query_nid, query_from, query_to, sbjct_nid, sbjct_from,
sbjct_to):
"""check, whether link is faithfull.
"""
query_profile = self.GetProfile(query_nid)
query_profile.useSegment(query_from, query_to)
sbjct_profile = self.GetProfile(sbjct_nid)
sbjct_profile.useSegment(sbjct_from, sbjct_to)
alignator = alignlib.makeFullDP(self.mGop, self.mGep)
result = alignlib.makeAlignataVector()
alignator.Align(query_profile, sbjct_profile, result)
if result.getLength() == 0:
query_profile.useFullLength()
sbjct_profile.useFullLength()
return 0, result
if self.mLogLevel >= 3:
print "# --> %i vs %i: score=%5.2f, length=%i, numgaps=%i, row_from=%i, row_to=%i, col_from=%i, col_to=%i" %\
(query_nid, sbjct_nid,
result.getScore(),
result.getLength(),
result.getNumGaps(),
result.getRowFrom(), result.getRowTo(),
result.getColFrom(), result.getColTo())
sys.stdout.flush()
if result.getScore() < self.mMinAlignmentScore:
query_profile.useFullLength()
sbjct_profile.useFullLength()
return 0, result
if result.getScore() > self.mSafetyThreshold * self.mMinAlignmentScore:
query_profile.useFullLength()
sbjct_profile.useFullLength()
return 1, result
z_params = alignlib.makeNormalDistributionParameters()
alignlib.calculateZScoreParameters(z_params, query_profile,
sbjct_profile, alignator,
self.mNumIterationsZScore)
mean = z_params.getMean()
stddev = z_params.getStandardDeviation()
if stddev == 0: stddev = 1
zscore = (result.getScore() - mean) / stddev
if self.mLogLevel >= 3:
print "# --> mean=%f, stdev=%f, zscore=%f" % (mean, stddev, zscore)
sys.stdout.flush()
query_profile.useFullLength()
sbjct_profile.useFullLength()
if zscore > self.mMinZScore:
return 1, result
else:
return 0, result
def buildMapPdb2Sequence( sequence, filename_pdb, options, pdb_chain = ""):
"""build a map for residue numbers in pdb file to residue numbers on
a sequence.
returns the following maps:
map_structure2seq: mapping of residue numbers between structure and
sequence. These are mappings that will work if you "renumber" the
structure.
map_pdb2seq, map_seq2pdb: mapping according to residue numbers in pdb file.
"""
if not os.path.exists( filename_pdb ):
return None, None
structure = Scientific.IO.PDB.Structure( filename_pdb )
map_pdb2seq = {}
map_seq2pdb = {}
for chain in structure.peptide_chains:
if chain.chain_id == pdb_chain:
## align pdb sequence to sequence
map_structure2seq = alignlib.makeAlignataVector()
alignator = alignlib.makeFullDP( -10.0, -2.0 )
## build sequence of pdb file
structure = ""
for residue in chain.sequence():
structure += AMINOACIDS[residue]
## align reference sequence to sequence of pdb file
row = alignlib.makeSequence( structure )
col = alignlib.makeSequence( sequence )
alignator.Align(row, col, map_structure2seq)
if options.loglevel >= 3:
options.stdlog.write( "structure: %s\n" % structure )
options.stdlog.write( "sequence : %s\n" % sequence )
options.stdlog.write( "alignment of structure to sequence:\n" )
options.stdlog.write( alignlib.writePairAlignment( row, col, map_structure2seq ) + "\n" )
# print alignlib.writeAlignataTable(map_structure2seq)
residue_number = 0
for residue in chain.residues:
residue_number += 1
mapped_residue = map_structure2seq.mapRowToCol(residue_number)
if not mapped_residue:
if options.loglevel >= 3:
options.stdlog.write( "# skipped residue %s=%s %i\n" % (str(residue.number), residue.name, residue_number))
continue
r = str(residue.number)
map_pdb2seq[r] = mapped_residue
map_seq2pdb[mapped_residue] = r
return map_structure2seq, map_pdb2seq, map_seq2pdb, residue_number-1, str(chain.residues[0].number), str(chain.residues[-1].number), structure
dbhandle = Picasso()
if not dbhandle.Connect():
print "Connection failed"
sys.exit(1)
x = AlignatorBenchmark( dbhandle )
x.SetTableSources( "picasso_benchmark.scop_sources" )
x.SetTableAlignments( "picasso_benchmark.scop_alignments" )
# if a sequence has repeated domains of the same type, only one gets chosen
statement = "SELECT DISTINCTROW rep_nid, rep_from, rep_to, SUBSTRING(scop_class, 1, 9) " +\
" FROM nrdb40_scop GROUP BY rep_nid, SUBSTRING(scop_class, 1, 9) "
x.FillSources( statement )
x.FilterSourcesDegeneracy( 2, 3 )
x.FilterSources( "SUBSTRING(class,1,3) > 4")
alignator = alignlib.makeFullDP( -12.0, -2.0 )
x.SetAlignator( alignator )
x.FillAlignments()
overlap, xcol_from, xcol_to,
(xcol_to - xcol_from) - (col_to - col_from)))
return ("0", ) * 12
##--------------------------------------------------------------------------------------
if __name__ == '__main__':
dbhandle = Pairsdb()
if not dbhandle.Connect():
print "Connection failed"
sys.exit(1)
a = alignlib.makeFullDP(-10.0, -2.0)
alignator = Alignator(a)
x = ScopTesterSequences(dbhandle, alignator)
x.Process()
if param_alignator == 0:
a = alignlib.makeFullDP(param_gop, param_gep)
alignator = Alignator(a)
if param_entities == 0:
tester = ScopTesterSequences(dbhandle, alignator)
tester.mLogLevel = param_loglevel
def CheckLink( self,
query_nid, query_from, query_to,
sbjct_nid, sbjct_from, sbjct_to):
"""check, whether link is faithfull.
"""
query_profile = self.GetProfile( query_nid )
query_profile.useSegment( query_from, query_to)
sbjct_profile = self.GetProfile( sbjct_nid )
sbjct_profile.useSegment( sbjct_from, sbjct_to)
alignator = alignlib.makeFullDP( self.mGop, self.mGep )
result = alignlib.makeAlignataVector()
alignator.Align( query_profile, sbjct_profile, result)
if result.getLength() == 0:
query_profile.useFullLength()
sbjct_profile.useFullLength()
return 0, result
if self.mLogLevel >= 3:
print "# --> %i vs %i: score=%5.2f, length=%i, numgaps=%i, row_from=%i, row_to=%i, col_from=%i, col_to=%i" %\
(query_nid, sbjct_nid,
result.getScore(),
result.getLength(),
result.getNumGaps(),
result.getRowFrom(), result.getRowTo(),
result.getColFrom(), result.getColTo())
sys.stdout.flush()
if result.getScore() < self.mMinAlignmentScore:
query_profile.useFullLength()
sbjct_profile.useFullLength()
return 0, result
if result.getScore() > self.mSafetyThreshold * self.mMinAlignmentScore:
query_profile.useFullLength()
sbjct_profile.useFullLength()
return 1,result
z_params = alignlib.makeNormalDistributionParameters()
alignlib.calculateZScoreParameters( z_params,
query_profile,
sbjct_profile,
alignator,
self.mNumIterationsZScore)
mean = z_params.getMean()
stddev = z_params.getStandardDeviation()
if stddev == 0: stddev = 1
zscore = (result.getScore() - mean) / stddev
if self.mLogLevel >= 3:
print "# --> mean=%f, stdev=%f, zscore=%f" % (mean, stddev, zscore)
sys.stdout.flush()
query_profile.useFullLength()
sbjct_profile.useFullLength()
if zscore > self.mMinZScore:
return 1,result
else:
return 0,result
