1.1 --- a/.hgignore	Fri Apr 29 17:01:29 2011 +0400
     1.2 +++ b/.hgignore	Fri Apr 29 17:02:41 2011 +0400
     1.3 @@ -26,8 +26,9 @@
     1.4  debian/python-allpy/
     1.5  debian/blocks3d/
     1.6  debian/blocks3d-wt/
     1.7 -debian/geometrical-core/
     1.8 +debian/geometrical-core2/
     1.9  
    1.10 +debian/bin/
    1.11  debian/*.substvars
    1.12  debian/*.debhelper*
    1.13  debian/*stamp*

     2.1 --- a/.hgtags	Fri Apr 29 17:01:29 2011 +0400
     2.2 +++ b/.hgtags	Fri Apr 29 17:02:41 2011 +0400
     2.3 @@ -1,2 +1,5 @@
     2.4  c51ef42eb5e5c2c98dac3c99271777905fb4da76 first run
     2.5  c1b67c10f8f3db62cce4b70c292a5882e91bf5b6 0.0
     2.6 +292b74612a42f2a1f270ad8184cc0883ad215564 1.1.0
     2.7 +364232e428887b5b7ec2f697c3aaa83c434449d3 1.2.0
     2.8 +b730512ecae51c89e8d268cd524270bbc50afc8a 1.3.0

     3.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
     3.2 +++ b/NEWS	Fri Apr 29 17:02:41 2011 +0400
     3.3 @@ -0,0 +1,4 @@
     3.4 +1.3.0 (2011-04-15)
     3.5 +
     3.6 +  * new: started NEWS file
     3.7 +  * new: started tracking version numbers

     4.1 --- a/README.devel	Fri Apr 29 17:01:29 2011 +0400
     4.2 +++ b/README.devel	Fri Apr 29 17:02:41 2011 +0400
     4.3 @@ -1,13 +1,42 @@
     4.4 +Using library in-place
     4.5 +======================
     4.6 +
     4.7  To use the library in-place, without installing, you have to add it's
     4.8  directory to PYTHONPATH.
     4.9  
    4.10 -e.g, type this in your shell:
    4.11 +e.g, type this in your shell::
    4.12  
    4.13 -	export PYTHONPATH=`pwd`
    4.14 +  export PYTHONPATH=`pwd`
    4.15  
    4.16 -alternatively, you can type:
    4.17 +alternatively, you can type::
    4.18  
    4.19    . ./SETPATH
    4.20  
    4.21  which adds the right path to PYTHONPATH and also adds 'utils' directory
    4.22 -to PATH
    4.23 +to PATH.
    4.24 +
    4.25 +NEWS file
    4.26 +=========
    4.27 +
    4.28 +NEWS file contains news that are important to the library users, not to the
    4.29 +devels. It must contain a word abot every change in any interfaces. It may
    4.30 +contain a word about bugfix, if that bugfix was important and could cause users
    4.31 +some grief. It must not contain anything else.
    4.32 +
    4.33 +The file grows upwards (like blog, so that all the most important things are at
    4.34 +the top, like)
    4.35 +
    4.36 +Example file format::
    4.37 +
    4.38 +  1.3.1 (2011-05-01)
    4.39 +    * bugfix: the Moon was going to fall
    4.40 +
    4.41 +  1.3.0 (2011-04-15)
    4.42 +
    4.43 +    * new: new feature description
    4.44 +    * new: new feature description
    4.45 +    * change(!): critical library interface change
    4.46 +    * change: change in the library interface
    4.47 +    * bugfix: short bug summary, see #ticket
    4.48 +    * bugfix: important bug that could destroy the Earth but did not hit trac
    4.49 +

     5.1 --- a/allpy/config.py	Fri Apr 29 17:01:29 2011 +0400
     5.2 +++ b/allpy/config.py	Fri Apr 29 17:02:41 2011 +0400
     5.3 @@ -23,6 +23,6 @@
     5.4  ac_count = 5
     5.5  
     5.6  # blocks3d
     5.7 -min_width = 3
     5.8 +min_width = 4
     5.9  timeout_2 = 10 # bron-kerbosh for blocks
    5.10  

     6.1 --- a/allpy/structure.py	Fri Apr 29 17:01:29 2011 +0400
     6.2 +++ b/allpy/structure.py	Fri Apr 29 17:02:41 2011 +0400
     6.3 @@ -316,6 +316,53 @@
     6.4                  columns.append(column)
     6.5          return columns
     6.6  
     6.7 +    def superimpose(self, gc, extra_columns=False):
     6.8 +        """ Superimpose monomers from this block at gc positions
     6.9 +
    6.10 +         * gc -- collection of columns to use as for superimposition
    6.11 +         * extra_columns -- allow columns with gaps or without structure
    6.12 +        """
    6.13 +        gc = list(gc)
    6.14 +        if len(self.sequences) >= 1:
    6.15 +            sequences = copy(self.sequences)
    6.16 +            main_sequence = sequences.pop()
    6.17 +            for sequence in sequences:
    6.18 +                fixed_gc = []
    6.19 +                moving_gc = []
    6.20 +                for column in gc:
    6.21 +                    if main_sequence in column and sequence in column:
    6.22 +                        if hasattr(column[main_sequence], 'pdb_residue') \
    6.23 +                        and hasattr(column[sequence], 'pdb_residue'):
    6.24 +                            fixed_gc.append(column[main_sequence].pdb_residue['CA'])
    6.25 +                            moving_gc.append(column[sequence].pdb_residue['CA'])
    6.26 +                        else:
    6.27 +                            assert extra_columns
    6.28 +                    else:
    6.29 +                        assert extra_columns
    6.30 +                sup = Superimposer()
    6.31 +                sup.set_atoms(fixed_gc, moving_gc)
    6.32 +                moving = sequence.pdb_chain.get_atoms()
    6.33 +                sup.apply(moving)
    6.34 +
    6.35 +    def save_pdb(self, out_file):
    6.36 +        """ Save all sequences
    6.37 +
    6.38 +        return {sequence: (new_chain, new_model)}
    6.39 +        """
    6.40 +        map = {}
    6.41 +        chains = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
    6.42 +        chain_index = -1
    6.43 +        model = 0
    6.44 +        for sequence in self.sequences:
    6.45 +            chain_index += 1
    6.46 +            if chain_index >= len(chains):
    6.47 +                chain_index = 0
    6.48 +                model += 1
    6.49 +            chain = chains[chain_index]
    6.50 +            sequence.save_pdb(out_file, chain, model)
    6.51 +            map[sequence] = (chain, model)
    6.52 +        return map
    6.53 +
    6.54  class BlockMixin(base.Block, AlignmentMixin):
    6.55      """Mixin to add 3D properties to blocks.
    6.56  
    6.57 @@ -576,39 +623,4 @@
    6.58                      result.append(column[sequence])
    6.59          return result
    6.60  
    6.61 -    def superimpose(self, gc):
    6.62 -        """ Superimpose monomers from this block at gc positions """
    6.63 -        gc = list(gc)
    6.64 -        if len(self.sequences) >= 1:
    6.65 -            sequences = copy(self.sequences)
    6.66 -            main_sequence = sequences.pop()
    6.67 -            fixed_gc = [column[main_sequence].pdb_residue['CA'] \
    6.68 -                for column in gc]
    6.69 -            for sequence in sequences:
    6.70 -                moving_gc = [column[sequence].pdb_residue['CA'] \
    6.71 -                    for column in gc]
    6.72 -                sup = Superimposer()
    6.73 -                sup.set_atoms(fixed_gc, moving_gc)
    6.74 -                moving = sequence.pdb_chain.get_atoms()
    6.75 -                sup.apply(moving)
    6.76 -
    6.77 -    def save_pdb(self, out_file):
    6.78 -        """ Save all sequences
    6.79 -
    6.80 -        return {sequence: (new_chain, new_model)}
    6.81 -        """
    6.82 -        map = {}
    6.83 -        chains = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
    6.84 -        chain_index = -1
    6.85 -        model = 0
    6.86 -        for sequence in self.sequences:
    6.87 -            chain_index += 1
    6.88 -            if chain_index >= len(chains):
    6.89 -                chain_index = 0
    6.90 -                model += 1
    6.91 -            chain = chains[chain_index]
    6.92 -            sequence.save_pdb(out_file, chain, model)
    6.93 -            map[sequence] = (chain, model)
    6.94 -        return map
    6.95 -
    6.96  # vim: set ts=4 sts=4 sw=4 et:

     7.1 --- a/debian/changelog	Fri Apr 29 17:01:29 2011 +0400
     7.2 +++ b/debian/changelog	Fri Apr 29 17:02:41 2011 +0400
     7.3 @@ -1,3 +1,27 @@
     7.4 +allpy (1.3.0-2) unstable; urgency=low
     7.5 +
     7.6 +  * renamed geometrical-core to geometrical-core2 by request of Boris,
     7.7 +    since the new version may yet be inferior to the old one in some aspects
     7.8 +
     7.9 + -- Danya Alexeyevsky <dendik@kodomo.fbb.msu.ru>  Fri, 15 Apr 2011 22:31:11 +0400
    7.10 +
    7.11 +allpy (1.3.0-1) unstable; urgency=low
    7.12 +
    7.13 +  * geometrical-core depends on particular version of python-allpy (>= 1.0.0)
    7.14 +
    7.15 + -- Danya Alexeyevsky <dendik@kodomo.fbb.msu.ru>  Fri, 15 Apr 2011 22:07:18 +0400
    7.16 +
    7.17 +allpy (1.3.0) unstable; urgency=low
    7.18 +
    7.19 +  * New allpy release.
    7.20 +    Version tracking an compact history starts with this version. The question
    7.21 +    of related programs and utils versioning is yet undecided.
    7.22 +  * New geometrical-core requires two packages with local install.
    7.23 +    Added sed script to add /usr/share/geometrical-core to sys.path.
    7.24 +    The required packages go there.
    7.25 +
    7.26 + -- Danya Alexeyevsky <dendik@kodomo.fbb.msu.ru>  Fri, 15 Apr 2011 21:21:40 +0400
    7.27 +
    7.28  allpy (0.0-6) unstable; urgency=medium
    7.29  
    7.30    * blocks3d-wt: reorganized apache configs for more stable operation

     8.1 --- a/debian/control.in	Fri Apr 29 17:01:29 2011 +0400
     8.2 +++ b/debian/control.in	Fri Apr 29 17:02:41 2011 +0400
     8.3 @@ -10,9 +10,9 @@
     8.4  Depends: ${python:Depends}, ${misc:Depends}
     8.5  Description: Python library for working with alignments
     8.6  
     8.7 -Package: geometrical-core
     8.8 +Package: geometrical-core2
     8.9  Architecture: all
    8.10 -Depends: ${python:Depends}, ${misc:Depends}
    8.11 +Depends: python-allpy (>= 1.0.0), ${python:Depends}, ${misc:Depends}
    8.12  Description: Find geometrical core in a multiple alignment
    8.13  
    8.14  Package: blocks3d

     9.1 --- a/debian/geometrical-core.install	Fri Apr 29 17:01:29 2011 +0400
     9.2 +++ /dev/null	Thu Jan 01 00:00:00 1970 +0000
     9.3 @@ -1,1 +0,0 @@
     9.4 -geometrical_core/geometrical-core usr/bin

    10.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
    10.2 +++ b/debian/geometrical-core2.install	Fri Apr 29 17:02:41 2011 +0400
    10.3 @@ -0,0 +1,2 @@
    10.4 +debian/bin/geometrical-core2 usr/bin
    10.5 +geometrical_core/*.py usr/share/geometrical-core2

    11.1 --- a/debian/rules	Fri Apr 29 17:01:29 2011 +0400
    11.2 +++ b/debian/rules	Fri Apr 29 17:02:41 2011 +0400
    11.3 @@ -11,6 +11,17 @@
    11.4  build/python-allpy::
    11.5  	make -C docs html
    11.6  
    11.7 +build/geometrical-core2::
    11.8 +	-mkdir debian/bin
    11.9 +	fix_path='import sys\nsys.path.append("/usr/share/geometrical-core2")'; \
   11.10 +	cat geometrical_core/geometrical-core \
   11.11 +		| sed "1 { x; s@^@\n$${fix_path}\n@; x }; /^$$/ x" \
   11.12 +		> debian/bin/geometrical-core2
   11.13 +	chmod +x debian/bin/geometrical-core2
   11.14 +
   11.15 +clean::
   11.16 +	rm -rf debian/bin
   11.17 +
   11.18  # Manually build blocks3d-wt to avoid build-dependency on qmake
   11.19  b3dwt = blocks3d/wt
   11.20  b3dwt_cpp = $(wildcard $(b3dwt)/*.C)

    12.1 --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
    12.2 +++ b/pair_cores/rasmol_output.py	Fri Apr 29 17:02:41 2011 +0400
    12.3 @@ -0,0 +1,85 @@
    12.4 +#!/usr/bin/python
    12.5 +
    12.6 +import os
    12.7 +import sys
    12.8 +from gzip import GzipFile
    12.9 +from copy import copy
   12.10 +
   12.11 +from protein_pdb import Alignment, Block, Monomer, Sequence
   12.12 +from allpy import structure
   12.13 +
   12.14 +def load(pdb_code):
   12.15 +    path = '/home/students/y08/bnagaev/pdb/%s/pdb%s.ent.gz' % (pdb_code[1:3], pdb_code)
   12.16 +    try:
   12.17 +        return GzipFile(path)
   12.18 +    except:
   12.19 +        return structure.download_pdb(pdb_code)
   12.20 +
   12.21 +def monomer2str(sequence, m):
   12.22 +    code = structure.pdb_id_parse(sequence.name)['code']
   12.23 +    model = structure.pdb_id_parse(sequence.name)['model']
   12.24 +    chain = structure.pdb_id_parse(sequence.name)['chain']
   12.25 +    resi = m.pdb_residue.get_id()[1]
   12.26 +    return '/%s/%i/%s/%i' % (code, model, chain, resi)
   12.27 +
   12.28 +def columns_for_align(columns, sequence, sequence2):
   12.29 +    cc = [c for c in columns if sequence in c and sequence2 in c and hasattr(c[sequence], 'pdb_residue') and hasattr(c[sequence2], 'pdb_residue')]
   12.30 +    str = ' or '.join(monomer2str(sequence, column[sequence]) for column in cc)
   12.31 +    return '((%s) and name ca)' % str
   12.32 +
   12.33 +aln_file = sys.argv[1]
   12.34 +blocks_file = sys.argv[2]
   12.35 +out_pdb = sys.argv[3]
   12.36 +
   12.37 +alignment = Alignment().append_file(open(aln_file))
   12.38 +blocks = alignment.blocks_from_file(open(blocks_file))
   12.39 +
   12.40 +for sequence in alignment.sequences:
   12.41 +    sequence.auto_pdb(pdb_getter=load)
   12.42 +
   12.43 +columns = set()
   12.44 +for b in blocks:
   12.45 +    columns |= set(b.columns)
   12.46 +columns = list(columns)
   12.47 +
   12.48 +_sequences = copy(alignment.sequences)
   12.49 +base_seq = _sequences.pop()
   12.50 +
   12.51 +alignment.superimpose(columns, extra_columns=True)
   12.52 +map = alignment.save_pdb(open(out_pdb, 'w'))
   12.53 +
   12.54 +for sequence, chain_model in map.items():
   12.55 +    print 'define str_%s *:%s' % (sequence.name, chain_model[0])
   12.56 +
   12.57 +print 'select all'
   12.58 +print 'color chain'
   12.59 +print 'restrict none'
   12.60 +print 'background [230, 240, 250]'
   12.61 +print 'select none'
   12.62 +
   12.63 +n = 0
   12.64 +
   12.65 +column2pos = alignment.column2pos()
   12.66 +
   12.67 +for block in blocks:
   12.68 +    selection = []
   12.69 +    for b in block.continuous_blocks():
   12.70 +        n += 1
   12.71 +        selections = []
   12.72 +        for sequence in b.sequences:
   12.73 +            try:
   12.74 +                chain = map[sequence][0]
   12.75 +                resi_from = b.columns[0][sequence].pdb_residue.get_id()[1]
   12.76 +                resi_to = b.columns[-1][sequence].pdb_residue.get_id()[1]
   12.77 +                selections.append('%i-%i:%s' % (resi_from, resi_to, chain))
   12.78 +            except:
   12.79 +                pass
   12.80 +        print ''
   12.81 +        print 'backbone'
   12.82 +        print 'define block_%i %s' % (n, ','.join(selections))
   12.83 +        print 'select block_%i' % n
   12.84 +        print 'echo block_%i %s %i-%i' % (n, ', '.join([b.sequences[0].name, b.sequences[1].name]),
   12.85 +            column2pos[b.columns[0]], column2pos[b.columns[-1]])
   12.86 +        print 'backbone 100'
   12.87 +        print 'pause'
   12.88 +

    13.1 --- a/pair_cores/score.py	Fri Apr 29 17:01:29 2011 +0400
    13.2 +++ b/pair_cores/score.py	Fri Apr 29 17:02:41 2011 +0400
    13.3 @@ -4,13 +4,15 @@
    13.4  import sys
    13.5  
    13.6  from allpy.graph import Graph
    13.7 +from allpy import config
    13.8  from protein_pdb import Alignment, Block, Monomer, Sequence
    13.9  
   13.10  def cn2(n):
   13.11      return n*(n-1)/2
   13.12  
   13.13 -def score(in_file, pair_cores_file, debug=False):
   13.14 +def score(in_file, pair_cores_file, debug=False, out_blocks=None):
   13.15      alignment = Alignment().append_file(open(in_file))
   13.16 +    assert os.path.exists(pair_cores_file) and os.path.getsize(pair_cores_file)
   13.17      pair_cores = alignment.blocks_from_file(open(pair_cores_file))
   13.18  
   13.19      C = []
   13.20 @@ -20,15 +22,27 @@
   13.21      for column in alignment.columns:
   13.22          column2blocks[column] = []
   13.23      for b in pair_cores:
   13.24 -        for column in b.columns[1:-1]:
   13.25 -            column2blocks[column].append(b)
   13.26 +        if len(b.columns) >= config.min_width:
   13.27 +            for column in b.columns[1:-1]:
   13.28 +                column2blocks[column].append(b)
   13.29 +    new_blocks = {}
   13.30      for column in alignment.columns:
   13.31          if len(column) >= 2:
   13.32              sequence_graph = Graph()
   13.33              for b in column2blocks[column]:
   13.34                  assert len(b.sequences) == 2
   13.35                  sequence_graph.set_edge(b.sequences[0], b.sequences[1])
   13.36 -            c = len(sequence_graph.connected_components())
   13.37 +            cc = sequence_graph.connected_components()
   13.38 +            c = len(cc)
   13.39 +            if out_blocks:
   13.40 +                for sequences in cc:
   13.41 +                    for s1 in sequences:
   13.42 +                        for s2 in sequences:
   13.43 +                            if s1 != s2:
   13.44 +                                new_blocks.setdefault(frozenset([s1,s2]),
   13.45 +                                    Block.from_alignment(alignment,
   13.46 +                                    columns=[],sequences=[s1,s2]))
   13.47 +                                new_blocks[frozenset([s1,s2])].columns.append(column)
   13.48              singles = len(column) - len(sequence_graph)
   13.49              c += singles
   13.50              C.append(c)
   13.51 @@ -36,15 +50,20 @@
   13.52              if debug:
   13.53                  print C[-1], L[-1]
   13.54  
   13.55 +    if out_blocks:
   13.56 +        alignment.blocks_to_file(open(out_blocks, 'w'), new_blocks.values())
   13.57 +
   13.58      score = sum([i*j for (i,j) in zip(C,L)]) / float(sum(L))
   13.59 -    print("%(score)f %(n)i %(name)s" % {'score': score,
   13.60 -        'n': len(alignment.sequences), 'name': in_file})
   13.61 +    print("%(score)f %(n)i %(columns)i %(name)s" % {'score': score,
   13.62 +        'n': len(alignment.sequences), 'name': in_file,
   13.63 +        'columns': len(L)})
   13.64  
   13.65  try:
   13.66      in_file = sys.argv[1]
   13.67      pair_cores_file = sys.argv[2]
   13.68      debug = len(sys.argv) >= 4 and sys.argv[3] == 'debug'
   13.69 -    score(in_file, pair_cores_file, debug)
   13.70 +    out_blocks = len(sys.argv) >= 5 and sys.argv[4]
   13.71 +    score(in_file, pair_cores_file, debug, out_blocks)
   13.72  except:
   13.73      pass
   13.74