IMP logo
IMP Reference Guide  2.6.0
The Integrative Modeling Platform
create_gmm.py
1 from __future__ import print_function
2 import IMP
3 import IMP.em
4 import IMP.isd
6 import numpy as np
7 from optparse import OptionParser
8 import sys,os
9 def parse_args():
10  usage = """%prog [options] <data_fn> <n_components> <out_txt_fn>
11  Create a GMM from either density file (.mrc), a pdb file (.pdb)
12  Will detect input format from extension.
13  Outputs as text and optionally as a density map
14  see help(-h)
15 """
16  parser = OptionParser(usage)
17 
18  parser.add_option("-t","--covar_type",dest="covar_type",default='full',
19  choices=['spherical', 'tied', 'diag', 'full'],
20  help="covariance type for the GMM")
21  parser.add_option("-m","--out_map",dest="out_map",default='',
22  help="write out the gmm to an mrc file")
23  parser.add_option("-a","--apix",dest="apix",default=1.0,type='float',
24  help="if you don't provide a map, set the voxel_size here (for sampling)")
25  parser.add_option("-n","--num_samples",dest="num_samples",default=1000000,type='int',
26  help="num samples to draw from the density map")
27  parser.add_option("-i","--num_iter",dest="num_iter",default=100,type='int',
28  help="num iterations of GMM")
29  parser.add_option("-s","--threshold",dest="threshold",default=0.0,type='float',
30  help="threshold for the map before sampling")
31 
32  parser.add_option("-c","--input_anchors_fn",dest="input_anchors_fn",default='',
33  help="get initial centers from anchors file. ")
34  parser.add_option("-f","--force_radii",dest="force_radii",default=-1.0,
35  type='float',
36  help="force radii to be this value (spherical) -1 means deactivated ")
37  parser.add_option("-w","--force_weight",dest="force_weight",default=-1.0,
38  type='float',
39  help="force weight to be this value (spherical) -1 means deactivated ")
40  parser.add_option("-e","--force_weight_frac",dest="force_weight_frac",action="store_true",default=False,
41  help="force weight to be 1.0/(num anchors). takes precedence over -w ")
42  parser.add_option("-o","--out_anchors_txt",dest="out_anchors_txt",default='',
43  help="write final GMM as anchor points (txt)")
44  parser.add_option("-q","--out_anchors_cmm",dest="out_anchors_cmm",default='',
45  help="write final GMM as anchor points (cmm)")
46  parser.add_option("-d","--use_dirichlet",dest="use_dirichlet",default=False,
47  action="store_true",
48  help="use dirichlet process for fit")
49 
50  parser.add_option("-k","--multiply_by_mass",dest="multiply_by_mass",default=False,
51  action="store_true",
52  help="if set, will multiply all weights by the total mass of the particles (PDB ONLY)")
53  parser.add_option("-x","--chain",dest="chain",default=None,
54  help="If you passed a PDB file, read this chain")
55 
56  parser.add_option("-z","--use_cpp",dest="use_cpp",default=False,
57  action="store_true",
58  help="EXPERIMENTAL. Uses the IMP GMM code. Requires isd_emxl")
59 
60 
61  (options, args) = parser.parse_args()
62  if len(args) != 3:
63  parser.error("incorrect number of arguments")
64  return options,args
65 
66 def run():
67  options,args = parse_args()
68  data_fn,ncenters,out_txt_fn = args
69  ncenters = int(ncenters)
70  mdl = IMP.Model()
71 
72  if not os.path.isfile(data_fn):
73  raise Exception("The data file you entered: "+data_fn+" does not exist!")
74 
75  ### get points for fitting the GMM
76  ext = data_fn.split('.')[-1]
77  mass_multiplier = 1.0
78  if ext=='pdb':
80  if options.chain:
81  mps = IMP.atom.Selection(mh,chain=options.chain).get_selected_particles()
82  else:
83  mps = IMP.core.get_leaves(mh)
84 
85  if options.multiply_by_mass:
86  mass_multiplier=sum(IMP.atom.Mass(p).get_mass() for p in mps)
87 
88  pts = [IMP.core.XYZ(p).get_coordinates() for p in mps]
89  bbox = None
90  elif ext=='mrc':
91  dmap = IMP.em.read_map(data_fn,IMP.em.MRCReaderWriter())
92  bbox = IMP.em.get_bounding_box(dmap)
93  print('sampling points')
94  pts = IMP.isd.sample_points_from_density(dmap,options.num_samples,options.threshold)
95  else:
96  print('ERROR: data_fn extension must be pdb, mrc, or npy')
97  sys.exit()
98 
99  ### Do fitting to points
100  if not options.use_cpp:
101  density_ps = []
102  print('fitting gmm')
103  #IMP.isd_emxl.gmm_tools.draw_points(pts,'test_points.bild')
104 
105  if options.force_weight_frac:
106  force_weight = 1.0/ncenters
107  else:
108  force_weight=options.force_weight
109  if force_weight != -1:
110  print('weight forced to',force_weight)
111  if not options.use_dirichlet:
112  gmm = IMP.isd.gmm_tools.fit_gmm_to_points(pts,ncenters,mdl,density_ps,
113  options.num_iter,options.covar_type,
114  force_radii=options.force_radii,
115  force_weight=options.force_weight,
116  mass_multiplier=mass_multiplier)
117  else:
118  gmm = IMP.isd.gmm_tools.fit_dirichlet_gmm_to_points(pts,ncenters,mdl,density_ps,
119  options.num_iter,options.covar_type,
120  mass_multiplier=mass_multiplier)
121 
122  else:
123  try:
124  import isd_emxl
125  except ImportError:
126  print("This option is experimental, only works if you have isd_emxl")
127  gmm_threshold = 0.01
128  density_ps = IMP.isd_emxl.fit_gaussians_to_density(mdl,dmap,options.num_samples,
129  ncenters,options.num_iter,
130  options.threshold,
131  gmm_threshold)
132 
133  ### Write to files
134  IMP.isd.gmm_tools.write_gmm_to_text(density_ps,out_txt_fn)
135  if options.out_map!='':
136  IMP.isd.gmm_tools.write_gmm_to_map(density_ps,options.out_map,
137  options.apix,
138  bbox)
139 
140  if options.out_anchors_txt!='':
141  IMP.isd.gmm_tools.write_gmm_to_anchors(density_ps,options.out_anchors_txt,
142  options.out_anchors_cmm)
143 
144 
145 
146 if __name__=="__main__":
147  run()
Select non water and non hydrogen atoms.
Definition: pdb.h:243
Tools for handling Gaussian Mixture Models.
Definition: gmm_tools.py:1
Add mass to a particle.
Definition: Mass.h:23
double get_mass(ResidueType c)
Get the mass from the residue type.
def fit_dirichlet_gmm_to_points
fit a GMM to some points.
Definition: gmm_tools.py:312
GenericHierarchies get_leaves(Hierarchy mhd)
Get all the leaves of the bit of hierarchy.
void read_pdb(TextInput input, int model, Hierarchy h)
Class for storing model, its restraints, constraints, and particles.
Definition: Model.h:72
def fit_gmm_to_points
fit a GMM to some points.
Definition: gmm_tools.py:224
Basic utilities for handling cryo-electron microscopy 3D density maps.
A decorator for a particle with x,y,z coordinates.
Definition: XYZ.h:30
def write_gmm_to_map
write density map from GMM.
Definition: gmm_tools.py:80
algebra::BoundingBoxD< 3 > get_bounding_box(const DensityMap *m)
Definition: DensityMap.h:457
def write_gmm_to_text
write a list of gaussians to text.
Definition: gmm_tools.py:62
Select hierarchy particles identified by the biological name.
Definition: Selection.h:66
Inferential scoring building on methods developed as part of the Inferential Structure Determination ...