| 1 |
""" |
| 2 |
ProMix: Take a protein structure ensemble and find a mixture of rigid |
| 3 |
segments or a mixture of conformers. Writes K copies of the ensemble |
| 4 |
(for segments) or K subsets of the ensemble (for conformers) as PDB |
| 5 |
files, each superposed on different components. |
| 6 |
|
| 7 |
Reference: Hirsch M, Habeck M. - Bioinformatics. 2008 Oct 1;24(19):2184-92 |
| 8 |
""" |
| 9 |
|
| 10 |
import numpy |
| 11 |
|
| 12 |
import csb.apps |
| 13 |
import csb.bio.structure |
| 14 |
|
| 15 |
from csb.bio.io.wwpdb import LegacyStructureParser |
| 16 |
from csb.statistics import mixtures |
| 17 |
|
| 18 |
|
| 19 |
class ExitCodes(csb.apps.ExitCodes): |
| 20 |
IO_ERROR = 2 |
| 21 |
|
| 22 |
class AppRunner(csb.apps.AppRunner): |
| 23 |
|
| 24 |
@property |
| 25 |
def target(self): |
| 26 |
return ProMixApp |
| 27 |
|
| 28 |
def command_line(self): |
| 29 |
cmd = csb.apps.ArgHandler(self.program, __doc__) |
| 30 |
|
| 31 |
cmd.add_scalar_option('components', 'K', int, 'Number of components', -1) |
| 32 |
cmd.add_scalar_option('type', 't', str, 'Type of mixture', 'segments', ('segments', 'conformers')) |
| 33 |
cmd.add_positional_argument('infile', str, 'input PDB file') |
| 34 |
|
| 35 |
return cmd |
| 36 |
|
| 37 |
def initapp(self, args): |
| 38 |
app = self.target |
| 39 |
return app(args) |
| 40 |
|
| 41 |
class ProMixApp(csb.apps.Application): |
| 42 |
|
| 43 |
def main(self): |
| 44 |
try: |
| 45 |
parser = LegacyStructureParser(self.args.infile) |
| 46 |
models = parser.models() |
| 47 |
except: |
| 48 |
self.exit('PDB file parsing failed', ExitCodes.IO_ERROR) |
| 49 |
|
| 50 |
if len(models) < 2: |
| 51 |
self.exit('PDB file contains only one model', ExitCodes.USAGE_ERROR) |
| 52 |
|
| 53 |
ensemble = parser.parse_models(models) |
| 54 |
X = numpy.array([model.get_coordinates(['CA'], True) for model in ensemble]) |
| 55 |
|
| 56 |
if self.args.type == 'segments': |
| 57 |
self.main_segments(ensemble, X) |
| 58 |
elif self.args.type == 'conformers': |
| 59 |
self.main_conformers(ensemble, X) |
| 60 |
else: |
| 61 |
raise ValueError('type must be "segments" or "conformers"') |
| 62 |
|
| 63 |
def main_segments(self, ensemble, X): |
| 64 |
|
| 65 |
mixture = mixtures.SegmentMixture.new(X, self.args.components) |
| 66 |
self.log('Number of segments: {0}'.format(mixture.K)) |
| 67 |
|
| 68 |
for k,(sigma,w) in enumerate(zip(mixture.sigma, mixture.w)): |
| 69 |
outfile = 'promix_segment_{0}.pdb'.format(k+1) |
| 70 |
self.log(' {0}: sigma = {1:6.3f}, w = {2:.3f}, file = {3}'.format(k+1, sigma, w, outfile)) |
| 71 |
|
| 72 |
for model, R, t in zip(ensemble, mixture.R, mixture.t): |
| 73 |
if k > 0: |
| 74 |
model.transform(R[k-1], t[k-1]) |
| 75 |
R = R[k].T |
| 76 |
t = -numpy.dot(R, t[k]) |
| 77 |
model.transform(R, t) |
| 78 |
|
| 79 |
ensemble.to_pdb(outfile) |
| 80 |
|
| 81 |
def main_conformers(self, ensemble, X): |
| 82 |
|
| 83 |
mixture = mixtures.ConformerMixture.new(X, self.args.components) |
| 84 |
self.log('Number of conformers: {0}'.format(mixture.K)) |
| 85 |
|
| 86 |
membership = mixture.membership |
| 87 |
|
| 88 |
for k,(sigma,w) in enumerate(zip(mixture.sigma, mixture.w)): |
| 89 |
outfile = 'promix_conformer_{0}.pdb'.format(k+1) |
| 90 |
self.log(' {0}: sigma = {1:6.3f}, w = {2:.3f}, file = {3}'.format(k+1, sigma, w, outfile)) |
| 91 |
|
| 92 |
ek = csb.bio.structure.Ensemble() |
| 93 |
|
| 94 |
for model, R, t, mk in zip(ensemble, mixture.R, mixture.t, membership): |
| 95 |
if mk != k: |
| 96 |
continue |
| 97 |
R = R[k].T |
| 98 |
t = -numpy.dot(R, t[k]) |
| 99 |
model.transform(R, t) |
| 100 |
ek.models.append(model) |
| 101 |
|
| 102 |
ek.to_pdb(outfile) |
| 103 |
|
| 104 |
|
| 105 |
def main(): |
| 106 |
AppRunner().run() |
| 107 |
|
| 108 |
|
| 109 |
if __name__ == '__main__': |
| 110 |
main() |