allpy: 62e38970ffa2 lib/sequence.py

allpy

view lib/sequence.py @ 111:62e38970ffa2

broken: lib::project::column -- new returning value sequence::pdb_chain_chain_add completed

author	boris <bnagaev@gmail.com>
date	Sat, 23 Oct 2010 12:57:07 +0400
parents	b43daeb9d767
children	32b9f4fadd35

line source

1 #!/usr/bin/python

2 from monomer import AminoAcidType

3 from Bio.PDB import PDBParser, CaPPBuilder

4 from allpy_pdb import Pdb

5 import project

6 import sys

9 cappbuilder = CaPPBuilder()

11 class Sequence(object):

12 """

13 Mandatory data:

14 * name -- str with the name of sequence

15 * description -- str with description of the sequence

16 * monomers -- list of monomer objects (aminoacids or nucleotides)

17 * pdb_chains -- list of Bio.PDB.Chain's

19 TODO: class Protein: container of multiple Sequences,

20 class ProteinMononer: container of multiple Monomers

21 """

22 def __init__(self, monomers=None, name='', description=""):

23 if not monomers:

24 monomers = []

25 self.name = name

26 self.description = description

27 self.monomers = monomers

28 self.pdb_chains = []

30 def __len__(self):

31 return len(self.monomers)

33 def __str__(self):

34 return ''.join([monomer.type.code1 for monomer in self.monomers])

36 def __eq__(self, other):

37 return len(self) == len(other) and \

38 all([a==b for a, b in zip(self.monomers, other.monomers)])

40 def __ne__(self, other):

41 return not (self == other)

43 def pdb_chain_add(self, pdb_file, pdb_id, pdb_chain, pdb_model=0):

44 """

45 Reads Pdb chain from file

46 and align each Monomer with PDB.Residue (TODO)

47 """

48 name = Pdb.std_id(pdb_id, pdb_chain, pdb_model)

49 chain = PDBParser().get_structure(name, pdb_file)[pdb_model][pdb_chain]

50 self.pdb_chains.append(chain)

51 pdb_sequence = Sequence.from_pdb_chain(chain)

52 alignment = project.Project.from_sequences(self, pdb_sequence)

53 alignment.muscle_align()

54 for monomer, pdb_monomer in alignment.column(sequence=pdb_sequence, original=self):

55 monomer.pdb_residue_add(chain, pdb_monomer.pdb_residues[chain])

56 #~ pdb_monomer.pdb_residue_add(self.pdb_chain,

57 # FIXME: align and map

59 @staticmethod

60 def from_str(fasta_str, name='', description='', monomer_kind=AminoAcidType):

61 monomers = [monomer_kind.from_code1(aa).instance() for aa in fasta_str]

62 return Sequence(monomers, name, description)

64 @staticmethod

65 def from_pdb_chain(chain):

66 """

67 chain is Bio.PDB.Chain

68 returns Sequence with Monomers with link to Bio.PDB.Residue

69 """

70 peptide = cappbuilder.build_peptides(chain)[0]

71 sequence = Sequence()

72 sequence.pdb_chains = [chain]

73 for ca_atom in peptide.get_ca_list():

74 residue = ca_atom.get_parent()

75 monomer = AminoAcidType.from_pdb_residue(residue).instance()

76 monomer.pdb_residue_add(chain, residue)

77 sequence.monomers.append(monomer)

78 return sequence