segmentmodel.py 125 KB
Newer Older
cmaffeo2's avatar
cmaffeo2 committed
1
import pdb
2
from pathlib import Path
3
import numpy as np
4
import random
5
6
7
from .model.arbdmodel import PointParticle, ParticleType, Group, ArbdModel
from .coords import rotationAboutAxis, quaternion_from_matrix, quaternion_to_matrix
from .model.nonbonded import *
8
from copy import copy, deepcopy
9
from .model.nbPot import nbDnaScheme
10

cmaffeo2's avatar
cmaffeo2 committed
11
12
from scipy.special import erf
import scipy.optimize as opt
13
from scipy import interpolate
cmaffeo2's avatar
cmaffeo2 committed
14

15
16
from .model.CanonicalNucleotideAtoms import canonicalNtFwd, canonicalNtRev, seqComplement
from .model.CanonicalNucleotideAtoms import enmTemplateHC, enmTemplateSQ, enmCorrectionsHC
cmaffeo2's avatar
cmaffeo2 committed
17

cmaffeo2's avatar
cmaffeo2 committed
18
19
from .model.spring_from_lp import k_angle as angle_spring_from_lp

20
# import pdb
21
"""
cmaffeo2's avatar
cmaffeo2 committed
22
TODO:
cmaffeo2's avatar
cmaffeo2 committed
23
 + fix handling of crossovers for atomic representation
cmaffeo2's avatar
cmaffeo2 committed
24
 + map to atomic representation
25
    + add nicks
cmaffeo2's avatar
cmaffeo2 committed
26
    + transform ssDNA nucleotides 
cmaffeo2's avatar
cmaffeo2 committed
27
    - shrink ssDNA
cmaffeo2's avatar
cmaffeo2 committed
28
    + shrink dsDNA backbone
29
    + make orientation continuous
cmaffeo2's avatar
cmaffeo2 committed
30
    + sequence
31
    + handle circular dna
32
 + ensure crossover bead potentials aren't applied twice 
33
 + remove performance bottlenecks
34
 - test for large systems
cmaffeo2's avatar
cmaffeo2 committed
35
 + assign sequence
36
 + ENM
37
38
 - rework Location class 
 - remove recursive calls
39
 - document
40
 - develop unit test suite
41
"""
cmaffeo2's avatar
cmaffeo2 committed
42
43
class CircularDnaError(Exception):
    pass
44

cmaffeo2's avatar
cmaffeo2 committed
45
46
47
class ParticleNotConnectedError(Exception):
    pass

48
49
class Location():
    """ Site for connection within an object """
50
    def __init__(self, container, address, type_, on_fwd_strand = True):
51
        ## TODO: remove cyclic references(?)
52
        self.container = container
cmaffeo2's avatar
cmaffeo2 committed
53
        self.address = address  # represents position along contour length in segment
cmaffeo2's avatar
cmaffeo2 committed
54
        # assert( type_ in ("end3","end5") ) # TODO remove or make conditional
55
        self.on_fwd_strand = on_fwd_strand
56
57
        self.type_ = type_
        self.particle = None
58
        self.connection = None
59
        self.is_3prime_side_of_connection = None
60

61
62
        self.prev_in_strand = None
        self.next_in_strand = None
63
64
        
        self.combine = None     # some locations might be combined in bead model 
65
66
67
68
69
70
71

    def get_connected_location(self):
        if self.connection is None:
            return None
        else:
            return self.connection.other(self)

72
    def set_connection(self, connection, is_3prime_side_of_connection):
73
        self.connection = connection # TODO weakref? 
74
        self.is_3prime_side_of_connection = is_3prime_side_of_connection
75

76
77
78
79
80
81
82
    def get_nt_pos(self):
        try:
            pos = self.container.contour_to_nt_pos(self.address, round_nt=True)
        except:
            if self.address == 0:
                pos = 0
            elif self.address == 1:
83
                pos = self.container.num_nt-1
84
85
86
87
            else:
                raise
        return pos

88
89
90
91
92
93
94
    def __repr__(self):
        if self.on_fwd_strand:
            on_fwd = "on_fwd_strand"
        else:
            on_fwd = "on_rev_strand"
        return "<Location {}.{}[{:.2f},{:d}]>".format( self.container.name, self.type_, self.address, self.on_fwd_strand)
        
95
96
97
98
99
100
101
102
103
class Connection():
    """ Abstract base class for connection between two elements """
    def __init__(self, A, B, type_ = None):
        assert( isinstance(A,Location) )
        assert( isinstance(B,Location) )
        self.A = A
        self.B = B
        self.type_ = type_
        
104
105
106
107
108
109
110
    def other(self, location):
        if location is self.A:
            return self.B
        elif location is self.B:
            return self.A
        else:
            raise Exception("OutOfBoundsError")
cmaffeo2's avatar
cmaffeo2 committed
111

112
113
    def delete(self):
        self.A.container.connections.remove(self)
114
115
        if self.B.container is not self.A.container:
            self.B.container.connections.remove(self)
116
117
118
        self.A.connection = None
        self.B.connection = None

cmaffeo2's avatar
cmaffeo2 committed
119
120
121
122
    def __repr__(self):
        return "<Connection {}--{}--{}]>".format( self.A, self.type_, self.B )
        

123
        
124
125
126
# class ConnectableElement(Transformable):
class ConnectableElement():
    """ Abstract base class """
cmaffeo2's avatar
cmaffeo2 committed
127
128
129
130
    def __init__(self, connection_locations=None, connections=None):
        if connection_locations is None: connection_locations = []
        if connections is None: connections = []

131
132
        ## TODO decide on names
        self.locations = self.connection_locations = connection_locations
133
134
        self.connections = connections

cmaffeo2's avatar
cmaffeo2 committed
135
    def get_locations(self, type_=None, exclude=()):
136
137
138
139
140
141
142
143
        locs = [l for l in self.connection_locations if (type_ is None or l.type_ == type_) and l.type_ not in exclude]
        counter = dict()
        for l in locs:
            if l in counter:
                counter[l] += 1
            else:
                counter[l] = 1
        assert( np.all( [counter[l] == 1 for l in locs] ) )
144
145
146
147
        return locs

    def get_location_at(self, address, on_fwd_strand=True, new_type="crossover"):
        loc = None
148
        if (self.num_nt == 1):
149
150
151
152
153
154
155
            # import pdb
            # pdb.set_trace()
            ## Assumes that intrahelical connections have been made before crossovers
            for l in self.locations:
                if l.on_fwd_strand == on_fwd_strand and l.connection is None:
                    assert(loc is None)
                    loc = l
cmaffeo2's avatar
cmaffeo2 committed
156
            # assert( loc is not None )
157
158
159
160
161
162
163
164
        else:
            for l in self.locations:
                if l.address == address and l.on_fwd_strand == on_fwd_strand:
                    assert(loc is None)
                    loc = l
        if loc is None:
            loc = Location( self, address=address, type_=new_type, on_fwd_strand=on_fwd_strand )
        return loc
165

cmaffeo2's avatar
cmaffeo2 committed
166
    def get_connections_and_locations(self, connection_type=None, exclude=()):
167
168
        """ Returns a list with each entry of the form:
            connection, location_in_self, location_in_other """
169
        type_ = connection_type
170
171
        ret = []
        for c in self.connections:
172
            if (type_ is None or c.type_ == type_) and c.type_ not in exclude:
173
                if   c.A.container is self:
174
                    ret.append( [c, c.A, c.B] )
175
                elif c.B.container is self:
176
177
                    ret.append( [c, c.B, c.A] )
                else:
178
179
                    import pdb
                    pdb.set_trace()
180
181
182
                    raise Exception("Object contains connection that fails to refer to object")
        return ret

183
    def _connect(self, other, connection, in_3prime_direction=None):
184
185
        ## TODO fix circular references        
        A,B = [connection.A, connection.B]
186
187
188
189
        if in_3prime_direction is not None:
            A.is_3prime_side_of_connection = not in_3prime_direction
            B.is_3prime_side_of_connection = in_3prime_direction
            
190
        A.connection = B.connection = connection
191
        self.connections.append(connection)
192
193
        if other is not self:
            other.connections.append(connection)
cmaffeo2's avatar
cmaffeo2 committed
194
195
        else:
            raise NotImplementedError("Segments cannot yet be connected to themselves; if you are attempting to make a circular object, try breaking the object into multiple segments")
196
197
198
199
200
201
        l = A.container.locations
        if A not in l: l.append(A)
        l = B.container.locations
        if B not in l: l.append(B)
        

202
203
    # def _find_connections(self, loc):
    #     return [c for c in self.connections if c.A == loc or c.B == loc]
204
205

class SegmentParticle(PointParticle):
cmaffeo2's avatar
cmaffeo2 committed
206
    def __init__(self, type_, position, name="A", **kwargs):
207
        self.name = name
208
        self.contour_position = None
cmaffeo2's avatar
cmaffeo2 committed
209
        PointParticle.__init__(self, type_, position, name=name, **kwargs)
210
211
        self.intrahelical_neighbors = []
        self.other_neighbors = []
cmaffeo2's avatar
cmaffeo2 committed
212
        self.locations = []
213
214
215

    def get_intrahelical_above(self):
        """ Returns bead directly above self """
216
        # assert( len(self.intrahelical_neighbors) <= 2 )
217
218
219
220
221
222
        for b in self.intrahelical_neighbors:
            if b.get_contour_position(self.parent) > self.contour_position:
                return b

    def get_intrahelical_below(self):
        """ Returns bead directly below self """
223
        # assert( len(self.intrahelical_neighbors) <= 2 )
224
225
226
227
        for b in self.intrahelical_neighbors:
            if b.get_contour_position(self.parent) < self.contour_position:
                return b

cmaffeo2's avatar
cmaffeo2 committed
228
    def _neighbor_should_be_added(self,b):
229
230
231
        if type(self.parent) != type(b.parent):
            return True

cmaffeo2's avatar
cmaffeo2 committed
232
233
234
235
        c1 = self.contour_position
        c2 = b.get_contour_position(self.parent)
        if c2 < c1:
            b0 = self.get_intrahelical_below()
236
        else:
cmaffeo2's avatar
cmaffeo2 committed
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
            b0 = self.get_intrahelical_above()

        if b0 is not None:            
            c0 = b0.get_contour_position(self.parent)
            if np.abs(c2-c1) < np.abs(c0-c1):
                ## remove b0
                self.intrahelical_neighbors.remove(b0)
                b0.intrahelical_neighbors.remove(self)
                return True
            else:
                return False
        return True
        
    def make_intrahelical_neighbor(self,b):
        add1 = self._neighbor_should_be_added(b)
        add2 = b._neighbor_should_be_added(self)
        if add1 and add2:
254
255
            # assert(len(b.intrahelical_neighbors) <= 1)
            # assert(len(self.intrahelical_neighbors) <= 1)
cmaffeo2's avatar
cmaffeo2 committed
256
257
            self.intrahelical_neighbors.append(b)
            b.intrahelical_neighbors.append(self)
258

cmaffeo2's avatar
cmaffeo2 committed
259
260
261
262
263
264
265
266
267
268
    def get_nt_position(self, seg):
        """ Returns the "address" of the nucleotide relative to seg in
        nucleotides, taking the shortest (intrahelical) contour length route to seg
        """
        if seg == self.parent:
            return seg.contour_to_nt_pos(self.contour_position)
        else:
            pos = self.get_contour_position(seg)
            return seg.contour_to_nt_pos(pos)

269
    def get_contour_position(self,seg):
cmaffeo2's avatar
cmaffeo2 committed
270
271
272
273
        """ TODO: fix paradigm where a bead maps to exactly one location in a polymer
        - One way: modify get_contour_position to take an optional argument that indicates where in the polymer you are looking from
        """

274
275
276
        if seg == self.parent:
            return self.contour_position
        else:
cmaffeo2's avatar
cmaffeo2 committed
277
278
279
280
281
            cutoff = 30*3
            target_seg = seg

            ## depth-first search
            ## TODO cache distances to nearby locations?
cmaffeo2's avatar
cmaffeo2 committed
282
            def descend_search_tree(seg, contour_in_seg, distance=0, visited_segs=None):
cmaffeo2's avatar
cmaffeo2 committed
283
                nonlocal cutoff
cmaffeo2's avatar
cmaffeo2 committed
284
                if visited_segs is None: visited_segs = []
cmaffeo2's avatar
cmaffeo2 committed
285
286
287
288

                if seg == target_seg:
                    # pdb.set_trace()
                    ## Found a segment in our target
289
290
                    sign = 1 if contour_in_seg == 1 else -1
                    if sign == -1: assert( contour_in_seg == 0 )
cmaffeo2's avatar
cmaffeo2 committed
291
292
                    if distance < cutoff: # TODO: check if this does anything
                        cutoff = distance
293
                    return [[distance, contour_in_seg+sign*seg.nt_pos_to_contour(distance)]], [(seg, contour_in_seg, distance)]
cmaffeo2's avatar
cmaffeo2 committed
294
                if distance > cutoff:
295
                    return None,None
cmaffeo2's avatar
cmaffeo2 committed
296
297
                    
                ret_list = []
298
                hist_list = []
cmaffeo2's avatar
cmaffeo2 committed
299
                ## Find intrahelical locations in seg that we might pass through
300
301
302
303
304
                conn_locs = seg.get_connections_and_locations("intrahelical")
                if isinstance(target_seg, SingleStrandedSegment):
                    tmp = seg.get_connections_and_locations("sscrossover")
                    conn_locs = conn_locs + list(filter(lambda x: x[2].container == target_seg, tmp))
                for c,A,B in conn_locs:
cmaffeo2's avatar
cmaffeo2 committed
305
                    if B.container in visited_segs: continue
306
307
308
                    dx = seg.contour_to_nt_pos( A.address, round_nt=False ) - seg.contour_to_nt_pos( contour_in_seg, round_nt=False)
                    dx = np.abs(dx)
                    results,history = descend_search_tree( B.container, B.address,
cmaffeo2's avatar
cmaffeo2 committed
309
310
311
                                                   distance+dx, visited_segs + [seg] )
                    if results is not None:
                        ret_list.extend( results )
312
313
                        hist_list.extend( history )
                return ret_list,hist_list
cmaffeo2's avatar
cmaffeo2 committed
314

315
            results,history = descend_search_tree(self.parent, self.contour_position)
cmaffeo2's avatar
cmaffeo2 committed
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
            if results is None or len(results) == 0:
                raise Exception("Could not find location in segment") # TODO better error
            return sorted(results,key=lambda x:x[0])[0][1]

            # nt_pos = self.get_nt_position(seg)
            # return seg.nt_pos_to_contour(nt_pos)

    def update_position(self, contour_position):
        self.contour_position = contour_position
        self.position = self.parent.contour_to_position(contour_position)
        if 'orientation_bead' in self.__dict__:
            o = self.orientation_bead
            o.contour_position = contour_position
            orientation = self.parent.contour_to_orientation(contour_position)
            if orientation is None:
                print("WARNING: local_twist is True, but orientation is None; using identity")
                orientation = np.eye(3)
            o.position = self.position + orientation.dot( np.array((Segment.orientation_bond.r0,0,0)) )
            
cmaffeo2's avatar
cmaffeo2 committed
335
336
337
    def __repr__(self):
        return "<SegmentParticle {} on {}[{:.2f}]>".format( self.name, self.parent, self.contour_position)

338
339

## TODO break this class into smaller, better encapsulated pieces
340
341
342
343
344
345
346
347
348
349
350
class Segment(ConnectableElement, Group):

    """ Base class that describes a segment of DNA. When built from
    cadnano models, should not span helices """

    """Define basic particle types"""
    dsDNA_particle = ParticleType("D",
                                  diffusivity = 43.5,
                                  mass = 300,
                                  radius = 3,                 
                              )
cmaffeo2's avatar
cmaffeo2 committed
351
352
353
354
355
    orientation_particle = ParticleType("O",
                                        diffusivity = 100,
                                        mass = 300,
                                        radius = 1,
                                    )
356

cmaffeo2's avatar
cmaffeo2 committed
357
    # orientation_bond = HarmonicBond(10,2)
358
    orientation_bond = HarmonicBond(30,1.5, rRange = (0,500) )
359
360
361
362
363
364
365

    ssDNA_particle = ParticleType("S",
                                  diffusivity = 43.5,
                                  mass = 150,
                                  radius = 3,                 
                              )

366
    def __init__(self, name, num_nt, 
cmaffeo2's avatar
cmaffeo2 committed
367
                 start_position = None,
368
                 end_position = None, 
cmaffeo2's avatar
cmaffeo2 committed
369
370
                 segment_model = None,
                 **kwargs):
371

cmaffeo2's avatar
cmaffeo2 committed
372
373
        if start_position is None: start_position = np.array((0,0,0))

cmaffeo2's avatar
cmaffeo2 committed
374
        Group.__init__(self, name, children=[], **kwargs)
375
        ConnectableElement.__init__(self, connection_locations=[], connections=[])
376

cmaffeo2's avatar
cmaffeo2 committed
377
378
379
        if 'segname' not in kwargs:
            self.segname = name
        # self.resname = name
cmaffeo2's avatar
cmaffeo2 committed
380
381
382
383
384
        self.start_orientation = None
        self.twist_per_nt = 0

        self.beads = [c for c in self.children] # self.beads will not contain orientation beads

385
386
387
        self._bead_model_generation = 0    # TODO: remove?
        self.segment_model = segment_model # TODO: remove?

388
389
390
391
        self.strand_pieces = dict()
        for d in ('fwd','rev'):
            self.strand_pieces[d] = []

392
        self.num_nt = int(num_nt)
393
        if end_position is None:
394
            end_position = np.array((0,0,self.distance_per_nt*num_nt)) + start_position
395
396
397
        self.start_position = start_position
        self.end_position = end_position

cmaffeo2's avatar
cmaffeo2 committed
398
399
400
401
        ## Used to assign cadnano names to beads
        self._generate_bead_callbacks = []
        self._generate_nucleotide_callbacks = []

402
        ## Set up interpolation for positions
403
404
405
406
        self._set_splines_from_ends()

        self.sequence = None

407
408
409
    def __repr__(self):
        return "<{} {}[{:d}]>".format( type(self), self.name, self.num_nt )

cmaffeo2's avatar
cmaffeo2 committed
410
    def set_splines(self, contours, coords):
411
        tck, u = interpolate.splprep( coords.T, u=contours, s=0, k=1)
412
        self.position_spline_params = (tck,u)
413

414
415
    def set_orientation_splines(self, contours, quaternions):
        tck, u = interpolate.splprep( quaternions.T, u=contours, s=0, k=1)
416
417
418
419
420
421
        self.quaternion_spline_params = (tck,u)

    def get_center(self):
        tck, u = self.position_spline_params
        return np.mean(self.contour_to_position(u), axis=0)

422
423
424
    def _get_location_positions(self):
        return [self.contour_to_nt_pos(l.address) for l in self.locations]

cmaffeo2's avatar
cmaffeo2 committed
425
    def insert_dna(self, at_nt: int, num_nt: int, seq=tuple()):
426
427
428
429
430
431
432
433
434
435
436
437
        assert(np.isclose(np.around(num_nt),num_nt))
        if at_nt < 0:
            raise ValueError("Attempted to insert DNA into {} at a negative location".format(self))
        if at_nt > self.num_nt-1:
            raise ValueError("Attempted to insert DNA into {} at beyond the end of the Segment".format(self))
        if num_nt < 0:
            raise ValueError("Attempted to insert DNA a negative amount of DNA into {}".format(self))

        num_nt = np.around(num_nt)
        nt_positions = self._get_location_positions()
        new_nt_positions = [p if p <= at_nt else p+num_nt for p in nt_positions]

cmaffeo2's avatar
cmaffeo2 committed
438
439
        ## TODO: handle sequence

440
441
442
443
444
445
        self.num_nt = self.num_nt+num_nt

        for l,p in zip(self.locations, new_nt_positions):
            l.address = self.nt_pos_to_contour(p)

    def remove_dna(self, first_nt: int, last_nt: int):
cmaffeo2's avatar
cmaffeo2 committed
446
        """ Removes nucleotides between first_nt and last_nt, inclusive """
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
        assert(np.isclose(np.around(first_nt),first_nt))
        assert(np.isclose(np.around(last_nt),last_nt))
        tmp = min((first_nt,last_nt))
        last_nt = max((first_nt,last_nt))
        fist_nt = tmp

        if first_nt < 0 or first_nt > self.num_nt-2:
            raise ValueError("Attempted to remove DNA from {} starting at an invalid location {}".format(self, first_nt))
        if last_nt < 1 or last_nt > self.num_nt-1:
            raise ValueError("Attempted to remove DNA from {} ending at an invalid location {}".format(self, last_nt))
        if first_nt == last_nt:
            return

        first_nt = np.around(first_nt)
        last_nt = np.around(last_nt)

        nt_positions = self._get_location_positions()

        bad_locations = list(filter(lambda p: p >= first_nt and p <= last_nt, nt_positions))
        if len(bad_locations) > 0:
            raise Exception("Attempted to remove DNA containing locations {} from {} between {} and {}".format(bad_locations,self,first_nt,last_nt))

cmaffeo2's avatar
cmaffeo2 committed
469
470
471
472
473
474
475
476
        removed_nt = last_nt-first_nt+1
        new_nt_positions = [p if p <= last_nt else p-removed_nt for p in nt_positions]
        num_nt = self.num_nt-removed_nt

        if self.sequence is not None and len(self.sequence) == self.num_nt:
            self.sequence = [s for s,i in zip(self.sequence,range(self.num_nt)) 
                                if i < first_nt or i > last_nt]
            assert( len(self.sequence) == num_nt )
477

cmaffeo2's avatar
cmaffeo2 committed
478
        self.num_nt = num_nt
479
480
481
482

        for l,p in zip(self.locations, new_nt_positions):
            l.address = self.nt_pos_to_contour(p)

483
484
485
486
487
488
489
490
491
492
493
494
    def __filter_contours(contours, positions, position_filter, contour_filter):
        u = contours
        r = positions

        ## Filter
        ids = list(range(len(u)))
        if contour_filter is not None:
            ids = list(filter(lambda i: contour_filter(u[i]), ids))
        if position_filter is not None:
            ids = list(filter(lambda i: position_filter(r[i,:]), ids))
        return ids

495
    def translate(self, translation_vector, position_filter=None, contour_filter=None):
496
497
498
499
500
        dr = np.array(translation_vector)
        tck, u = self.position_spline_params
        r = self.contour_to_position(u)

        ids = Segment.__filter_contours(u, r, position_filter, contour_filter)
cmaffeo2's avatar
cmaffeo2 committed
501
        if len(ids) == 0: return
502
503
504

        ## Translate
        r[ids,:] = r[ids,:] + dr[np.newaxis,:]
505
        self.set_splines(u,r)
506
507
508
509
510
511

    def rotate(self, rotation_matrix, about=None, position_filter=None, contour_filter=None):
        tck, u = self.position_spline_params
        r = self.contour_to_position(u)

        ids = Segment.__filter_contours(u, r, position_filter, contour_filter)
cmaffeo2's avatar
cmaffeo2 committed
512
        if len(ids) == 0: return
513
514
515

        if about is None:
            ## TODO: do this more efficiently
516
            r[ids,:] = np.array([rotation_matrix.dot(r[i,:]) for i in ids])
517
518
519
        else:
            dr = np.array(about)
            ## TODO: do this more efficiently
520
            r[ids,:] = np.array([rotation_matrix.dot(r[i,:]-dr) + dr for i in ids])
521

522
        self.set_splines(u,r)
523
524
525
526

        if self.quaternion_spline_params is not None:
            ## TODO: performance: don't shift between quaternion and matrix representations so much
            tck, u = self.quaternion_spline_params
527
            orientations = [self.contour_to_orientation(v) for v in u]
528
529
530
            for i in ids:
                orientations[i,:] = rotation_matrix.dot(orientations[i])
            quats = [quaternion_from_matrix(o) for o in orientations]
531
            self.set_orientation_splines(u, quats)
532

533
    def _set_splines_from_ends(self, resolution=4):
534
        self.quaternion_spline_params = None
535
536
537
538
539
540
        r0 = np.array(self.start_position)[np.newaxis,:]
        r1 = np.array(self.end_position)[np.newaxis,:]
        u = np.linspace(0,1, max(3,self.num_nt//int(resolution)))
        s = u[:,np.newaxis]
        coords = (1-s)*r0 + s*r1
        self.set_splines(u, coords)
541

542
543
544
    def clear_all(self):
        Group.clear_all(self)  # TODO: use super?
        self.beads = []
545
546
547
548
549
        # for c,loc,other in self.get_connections_and_locations():
        #     loc.particle = None
        #     other.particle = None
        for l in self.locations:
            l.particle = None
550

551
    def contour_to_nt_pos(self, contour_pos, round_nt=False):
552
        nt = contour_pos*(self.num_nt) - 0.5
553
        if round_nt:
cmaffeo2's avatar
cmaffeo2 committed
554
            assert( np.isclose(np.around(nt),nt) )
555
556
557
            nt = np.around(nt)
        return nt

558
    def nt_pos_to_contour(self,nt_pos):
559
        return (nt_pos+0.5)/(self.num_nt)
560

561
    def contour_to_position(self,s):
562
        p = interpolate.splev( s, self.position_spline_params[0] )
563
564
565
566
        if len(p) > 1: p = np.array(p).T
        return p

    def contour_to_tangent(self,s):
567
        t = interpolate.splev( s, self.position_spline_params[0], der=1 )
568
569
        t = (t / np.linalg.norm(t,axis=0))
        return t.T
570
571
572
        

    def contour_to_orientation(self,s):
573
        assert( isinstance(s,float) or isinstance(s,int) or len(s) == 1 )   # TODO make vectorized version
574

575
576
577
578
579
        if self.quaternion_spline_params is None:
            axis = self.contour_to_tangent(s)
            axis = axis / np.linalg.norm(axis)
            rotAxis = np.cross(axis,np.array((0,0,1)))
            rotAxisL = np.linalg.norm(rotAxis)
580
            zAxis = np.array((0,0,1))
581
582

            if rotAxisL > 0.001:
583
584
585
                theta = np.arcsin(rotAxisL) * 180/np.pi
                if axis.dot(zAxis) < 0: theta = 180-theta
                orientation0 = rotationAboutAxis( rotAxis/rotAxisL, theta, normalizeAxis=False ).T
586
            else:
587
588
                orientation0 = np.eye(3) if axis.dot(zAxis) > 0 else \
                               rotationAboutAxis( np.array((1,0,0)), 180, normalizeAxis=False )
589
590
591
            if self.start_orientation is not None:
                orientation0 = orientation0.dot(self.start_orientation)

592
593
594
            orientation = rotationAboutAxis( axis, self.twist_per_nt*self.contour_to_nt_pos(s), normalizeAxis=False )
            orientation = orientation.dot(orientation0)
        else:
595
            q = interpolate.splev( s, self.quaternion_spline_params[0] )
596
597
            if len(q) > 1: q = np.array(q).T # TODO: is this needed?
            orientation = quaternion_to_matrix(q)
598

599
        return orientation
600

cmaffeo2's avatar
cmaffeo2 committed
601
    def get_contour_sorted_connections_and_locations(self,type_):
cmaffeo2's avatar
cmaffeo2 committed
602
        sort_fn = lambda c: c[1].address
cmaffeo2's avatar
cmaffeo2 committed
603
        cl = self.get_connections_and_locations(type_)
cmaffeo2's avatar
cmaffeo2 committed
604
        return sorted(cl, key=sort_fn)
605
606
607
    
    def randomize_unset_sequence(self):
        bases = list(seqComplement.keys())
608
        # bases = ['T']        ## FOR DEBUG
609
        if self.sequence is None:
610
            self.sequence = [random.choice(bases) for i in range(self.num_nt)]
611
        else:
612
            assert(len(self.sequence) == self.num_nt) # TODO move
613
614
615
            for i in range(len(self.sequence)):
                if self.sequence[i] is None:
                    self.sequence[i] = random.choice(bases)
616

cmaffeo2's avatar
cmaffeo2 committed
617
618
619
    def _get_num_beads(self, max_basepairs_per_bead, max_nucleotides_per_bead ):
        raise NotImplementedError

620
    def _generate_one_bead(self, contour_position, nts):
621
622
        raise NotImplementedError

cmaffeo2's avatar
cmaffeo2 committed
623
    def _generate_atomic_nucleotide(self, contour_position, is_fwd, seq, scale, strand_segment):
cmaffeo2's avatar
cmaffeo2 committed
624
625
626
627
628
        """ Seq should include modifications like 5T, T3 Tsinglet; direction matters too """

        # print("Generating nucleotide at {}".format(contour_position))
        
        pos = self.contour_to_position(contour_position)
629
        orientation = self.contour_to_orientation(contour_position)
cmaffeo2's avatar
cmaffeo2 committed
630

631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
        """ deleteme
        ## TODO: move this code (?)
        if orientation is None:
            import pdb
            pdb.set_trace()
            axis = self.contour_to_tangent(contour_position)
            angleVec = np.array([1,0,0])
            if axis.dot(angleVec) > 0.9: angleVec = np.array([0,1,0])
            angleVec = angleVec - angleVec.dot(axis)*axis
            angleVec = angleVec/np.linalg.norm(angleVec)
            y = np.cross(axis,angleVec)
            orientation = np.array([angleVec,y,axis]).T
            ## TODO: improve placement of ssDNA
            # rot = rotationAboutAxis( axis, contour_position*self.twist_per_nt*self.num_nt, normalizeAxis=True )
            # orientation = rot.dot(orientation)
        else:
            orientation = orientation                            
        """
cmaffeo2's avatar
cmaffeo2 committed
649
        key = seq
650
651
        nt_dict = canonicalNtFwd if is_fwd else canonicalNtRev

652
        atoms = nt_dict[ key ].generate() # TODO: clone?
cmaffeo2's avatar
cmaffeo2 committed
653
        atoms.orientation = orientation.dot(atoms.orientation)
654
655
656
657
        if isinstance(self, SingleStrandedSegment):
            if scale is not None and scale != 1:
                for a in atoms:
                    a.position = scale*a.position
658
            atoms.position = pos - atoms.atoms_by_name["C1'"].collapsedPosition()
659
660
661
662
663
664
665
666
667
        else:
            if scale is not None and scale != 1:
                if atoms.sequence in ("A","G"):
                    r0 = atoms.atoms_by_name["N9"].position
                else:
                    r0 = atoms.atoms_by_name["N1"].position
                for a in atoms:
                    if a.name[-1] in ("'","P","T"):
                        a.position = scale*(a.position-r0) + r0
668
                    else:
669
                        a.fixed = 1
670
            atoms.position = pos
cmaffeo2's avatar
cmaffeo2 committed
671
672
673
674
675
676
        
        atoms.contour_position = contour_position
        strand_segment.add(atoms)

        for callback in self._generate_nucleotide_callbacks:
            callback(atoms)
cmaffeo2's avatar
cmaffeo2 committed
677
678

        return atoms
679

680
681
    def add_location(self, nt, type_, on_fwd_strand=True):
        ## Create location if needed, add to segment
682
        c = self.nt_pos_to_contour(nt)
683
684
685
686
687
688
        assert(c >= 0 and c <= 1)
        # TODO? loc = self.Location( address=c, type_=type_, on_fwd_strand=is_fwd )
        loc = Location( self, address=c, type_=type_, on_fwd_strand=on_fwd_strand )
        self.locations.append(loc)

    ## TODO? Replace with abstract strand-based model?
689
690
691
692
693

    def add_nick(self, nt, on_fwd_strand=True):
        self.add_3prime(nt,on_fwd_strand)
        self.add_5prime(nt+1,on_fwd_strand)

694
    def add_5prime(self, nt, on_fwd_strand=True):
695
696
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
697
        self.add_location(nt,"5prime",on_fwd_strand)
698
699

    def add_3prime(self, nt, on_fwd_strand=True):
700
701
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
702
        self.add_location(nt,"3prime",on_fwd_strand)
703

704
    def get_3prime_locations(self):
cmaffeo2's avatar
cmaffeo2 committed
705
        return sorted(self.get_locations("3prime"),key=lambda x: x.address)
706
    
cmaffeo2's avatar
cmaffeo2 committed
707
    def get_5prime_locations(self):
708
        ## TODO? ensure that data is consistent before _build_model calls
cmaffeo2's avatar
cmaffeo2 committed
709
        return sorted(self.get_locations("5prime"),key=lambda x: x.address)
cmaffeo2's avatar
cmaffeo2 committed
710

711
    def iterate_connections_and_locations(self, reverse=False):
cmaffeo2's avatar
cmaffeo2 committed
712
713
        ## connections to other segments
        cl = self.get_contour_sorted_connections_and_locations()
714
        if reverse:
cmaffeo2's avatar
cmaffeo2 committed
715
            cl = cl[::-1]
716
717
718
            
        for c in cl:
            yield c
cmaffeo2's avatar
cmaffeo2 committed
719

720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
    ## TODO rename
    def _add_strand_piece(self, strand_piece):
        """ Registers a strand segment within this object """

        ## TODO use weakref
        d = 'fwd' if strand_piece.is_fwd else 'rev'

        ## Validate strand_piece (ensure no clashes)
        for s in self.strand_pieces[d]:
            l,h = sorted((s.start,s.end))
            for value in (strand_piece.start,strand_piece.end):
                assert( value < l or value > h )

        ## Add strand_piece in correct order
        self.strand_pieces[d].append(strand_piece)
        self.strand_pieces[d] = sorted(self.strand_pieces[d],
                                       key = lambda x: x.start)

738
    ## TODO rename
739
    def get_strand_segment(self, nt_pos, is_fwd, move_at_least=0.5):
740
        """ Walks through locations, checking for crossovers """
741
742
743
744
        # if self.name in ("6-1","1-1"):
        #     import pdb
        #     pdb.set_trace()
        move_at_least = 0
745
746

        ## Iterate through locations
cmaffeo2's avatar
cmaffeo2 committed
747
        # locations = sorted(self.locations, key=lambda l:(l.address,not l.on_fwd_strand), reverse=(not is_fwd))
748
749
750
751
752
753
        def loc_rank(l):
            nt = l.get_nt_pos()
            ## optionally add logic about type of connection
            return (nt, not l.on_fwd_strand)
        # locations = sorted(self.locations, key=lambda l:(l.address,not l.on_fwd_strand), reverse=(not is_fwd))
        locations = sorted(self.locations, key=loc_rank, reverse=(not is_fwd))
754
755
        # print(locations)

756
        for l in locations:
cmaffeo2's avatar
cmaffeo2 committed
757
758
759
760
            # TODOTODO probably okay
            if l.address == 0:
                pos = 0.0
            elif l.address == 1:
761
                pos = self.num_nt-1
cmaffeo2's avatar
cmaffeo2 committed
762
763
            else:
                pos = self.contour_to_nt_pos(l.address, round_nt=True)
764
765
766

            ## DEBUG

cmaffeo2's avatar
cmaffeo2 committed
767

768
            ## Skip locations encountered before our strand
769
770
771
772
773
774
775
776
            # tol = 0.1
            # if is_fwd:
            #     if pos-nt_pos <= tol: continue 
            # elif   nt_pos-pos <= tol: continue
            if (pos-nt_pos)*(2*is_fwd-1) < move_at_least: continue
            ## TODO: remove move_at_least
            if np.isclose(pos,nt_pos):
                if l.is_3prime_side_of_connection: continue
777
778

            ## Stop if we found the 3prime end
779
            if l.on_fwd_strand == is_fwd and l.type_ == "3prime" and l.connection is None:
780
                # print("  found end at",l)
781
                return pos, None, None, None, None
782
783
784
785
786
787
788
789

            ## Check location connections
            c = l.connection
            if c is None: continue
            B = c.other(l)            

            ## Found a location on the same strand?
            if l.on_fwd_strand == is_fwd:
790
791
                # print("  passing through",l)
                # print("from {}, connection {} to {}".format(nt_pos,l,B))
792
                Bpos = B.get_nt_pos()
793
                return pos, B.container, Bpos, B.on_fwd_strand, 0.5
794
795
796
                
            ## Stop at other strand crossovers so basepairs line up
            elif c.type_ == "crossover":
797
                if nt_pos == pos: continue
798
                # print("  pausing at",l)
799
                return pos, l.container, pos+(2*is_fwd-1), is_fwd, 0
800
801
802
803
804
805

        raise Exception("Shouldn't be here")
        # print("Shouldn't be here")
        ## Made it to the end of the segment without finding a connection
        return 1*is_fwd, None, None, None

806
807
808
    def get_nearest_bead(self, contour_position):
        if len(self.beads) < 1: return None
        cs = np.array([b.contour_position for b in self.beads]) # TODO: cache
809
        # TODO: include beads in connections?
810
811
812
        i = np.argmin((cs - contour_position)**2)

        return self.beads[i]
813

814
815
816
817
818
819
820
821
822
    def _get_atomic_nucleotide(self, nucleotide_idx, is_fwd=True):
        d = 'fwd' if is_fwd else 'rev'
        for s in self.strand_pieces[d]:
            try:
                return s.get_nucleotide(nucleotide_idx)
            except:
                pass
        raise Exception("Could not find nucleotide in {} at {}.{}".format( self, nucleotide_idx, d ))

823
824
    def get_all_consecutive_beads(self, number):
        assert(number >= 1)
cmaffeo2's avatar
cmaffeo2 committed
825
        ## Assume that consecutive beads in self.beads are bonded
826
        ret = []
cmaffeo2's avatar
cmaffeo2 committed
827
828
        for i in range(len(self.beads)-number+1):
            tmp = [self.beads[i+j] for j in range(0,number)]
829
            ret.append( tmp )
830
        return ret   
831

832
833
834
    def _add_bead(self,b,set_contour=False):
        if set_contour:
            b.contour_position = b.get_contour_position(self)
835
        
836
837
838
        # assert(b.parent is None)
        if b.parent is not None:
            b.parent.children.remove(b)
839
        self.add(b)
840
841
842
843
844
845
        self.beads.append(b) # don't add orientation bead
        if "orientation_bead" in b.__dict__: # TODO: think of a cleaner approach
            o = b.orientation_bead
            o.contour_position = b.contour_position
            if o.parent is not None:
                o.parent.children.remove(o)
846
            self.add(o)
847
848
849
850
851
852
853
854
            self.add_bond(b,o, Segment.orientation_bond, exclude=True)

    def _rebuild_children(self, new_children):
        # print("_rebuild_children on %s" % self.name)
        old_children = self.children
        old_beads = self.beads
        self.children = []
        self.beads = []
855
856

        if True:
857
858
            ## TODO: remove this if duplicates are never found 
            # print("Searching for duplicate particles...")
859
            ## Remove duplicates, preserving order
860
861
862
863
            tmp = []
            for c in new_children:
                if c not in tmp:
                    tmp.append(c)
864
                else:
865
                    print("  DUPLICATE PARTICLE FOUND!")
866
867
            new_children = tmp

868
869
870
871
872
        for b in new_children:
            self.beads.append(b)
            self.children.append(b)
            if "orientation_bead" in b.__dict__: # TODO: think of a cleaner approach
                self.children.append(b.orientation_bead)
873
874
875
876
877
            
        # tmp = [c for c in self.children if c not in old_children]
        # assert(len(tmp) == 0)
        # tmp = [c for c in old_children if c not in self.children]
        # assert(len(tmp) == 0)
878
879
        assert(len(old_children) == len(self.children))
        assert(len(old_beads) == len(self.beads))
880

881

cmaffeo2's avatar
cmaffeo2 committed
882
    def _generate_beads(self, bead_model, max_basepairs_per_bead, max_nucleotides_per_bead):
883

884
        """ Generate beads (positions, types, etc) and bonds, angles, dihedrals, exclusions """
cmaffeo2's avatar
cmaffeo2 committed
885
        ## TODO: decide whether to remove bead_model argument
886
        ##       (currently unused)
cmaffeo2's avatar
cmaffeo2 committed
887

888
        ## First find points between-which beads must be generated
889
890
891
        # conn_locs = self.get_contour_sorted_connections_and_locations()
        # locs = [A for c,A,B in conn_locs]
        # existing_beads = [l.particle for l in locs if l.particle is not None]
cmaffeo2's avatar
cmaffeo2 committed
892
893
        # if self.name == "S001":
        #     pdb.set_trace()
894

cmaffeo2's avatar
cmaffeo2 committed
895
896
        existing_beads0 = {l.particle for l in self.locations if l.particle is not None}
        existing_beads = sorted( list(existing_beads0), key=lambda b: b.get_contour_position(self) )
897

898
899
900
        # if self.num_nt == 1 and all([l.particle is not None for l in self.locations]):
        #     pdb.set_trace()
        #     return
901

902
903
904
905
        for b in existing_beads:
            assert(b.parent is not None)

        ## Add ends if they don't exist yet
906
        ## TODOTODO: test 1 nt segments?
907
        if len(existing_beads) == 0 or existing_beads[0].get_nt_position(self) >= 0.5:
cmaffeo2's avatar
cmaffeo2 committed
908
909
            # if len(existing_beads) > 0:            
            #     assert(existing_beads[0].get_nt_position(self) >= 0.5)
910
            b = self._generate_one_bead( self.nt_pos_to_contour(0), 0)
911
            existing_beads = [b] + existing_beads
cmaffeo2's avatar
cmaffeo2 committed
912

cmaffeo2's avatar
cmaffeo2 committed
913
        if existing_beads[-1].get_nt_position(self)-(self.num_nt-1) < -0.5 or len(existing_beads)==1:
914
            b = self._generate_one_bead( self.nt_pos_to_contour(self.num_nt-1), 0)
915
916
917
918
919
            existing_beads.append(b)
        assert(len(existing_beads) > 1)

        ## Walk through existing_beads, add beads between
        tmp_children = []       # build list of children in nice order
920
        last = None
cmaffeo2's avatar
cmaffeo2 committed
921

922
923
        for I in range(len(existing_beads)-1):
            eb1,eb2 = [existing_beads[i] for i in (I,I+1)]
924
            assert( eb1 is not eb2 )
925

cmaffeo2's avatar
cmaffeo2 committed
926
927
928
929
            # if np.isclose(eb1.position[2], eb2.position[2]):
            #     import pdb
            #     pdb.set_trace()

930
            # print(" %s working on %d to %d" % (self.name, eb1.position[2], eb2.position[2]))
931
932
            e_ds = eb2.get_contour_position(self) - eb1.get_contour_position(self)
            num_beads = self._get_num_beads( e_ds, max_basepairs_per_bead, max_nucleotides_per_bead )
cmaffeo2's avatar
cmaffeo2 committed
933
934
935
936
937
938

            ## Ensure there is a ssDNA bead between dsDNA beads
            if num_beads == 0 and isinstance(self,SingleStrandedSegment) and isinstance(eb1.parent,DoubleStrandedSegment) and isinstance(eb2.parent,DoubleStrandedSegment):
                num_beads = 1
            ## TODO similarly ensure there is a dsDNA bead between ssDNA beads

939
            ds = e_ds / (num_beads+1)
940
941
942
            nts = ds*self.num_nt
            eb1.num_nt += 0.5*nts
            eb2.num_nt += 0.5*nts
943
944
945
946
947
948
949

            ## Add beads
            if eb1.parent == self:
                tmp_children.append(eb1)

            s0 = eb1.get_contour_position(self)
            if last is not None:
cmaffeo2's avatar
cmaffeo2 committed
950
                last.make_intrahelical_neighbor(eb1)
951
952
953
            last = eb1
            for j in range(num_beads):
                s = ds*(j+1) + s0
954
                # if self.name in ("51-2","51-3"):
cmaffeo2's avatar
cmaffeo2 committed
955
                # if self.name in ("31-2",):
956
                #     print(" adding bead at {}".format(s))
957
958
                b = self._generate_one_bead(s,nts)

cmaffeo2's avatar
cmaffeo2 committed
959
                last.make_intrahelical_neighbor(b)
960
961
962
                last = b
                tmp_children.append(b)

cmaffeo2's avatar
cmaffeo2 committed
963
        last.make_intrahelical_neighbor(eb2)
964
965
966

        if eb2.parent == self:
            tmp_children.append(eb2)
cmaffeo2's avatar
cmaffeo2 committed
967
        # if self.name in ("31-2",):
968
        #     pdb.set_trace()
969
        self._rebuild_children(tmp_children)
970

cmaffeo2's avatar
cmaffeo2 committed
971
972
973
        for callback in self._gen