segmentmodel.py 103 KB
Newer Older
cmaffeo2's avatar
cmaffeo2 committed
1
import pdb
2
import numpy as np
3
import random
4
5
6
from .model.arbdmodel import PointParticle, ParticleType, Group, ArbdModel
from .coords import rotationAboutAxis, quaternion_from_matrix, quaternion_to_matrix
from .model.nonbonded import *
7
from copy import copy, deepcopy
8
from .model.nbPot import nbDnaScheme
9

cmaffeo2's avatar
cmaffeo2 committed
10
11
from scipy.special import erf
import scipy.optimize as opt
12
from scipy import interpolate
cmaffeo2's avatar
cmaffeo2 committed
13

14
15
from .model.CanonicalNucleotideAtoms import canonicalNtFwd, canonicalNtRev, seqComplement
from .model.CanonicalNucleotideAtoms import enmTemplateHC, enmTemplateSQ, enmCorrectionsHC
cmaffeo2's avatar
cmaffeo2 committed
16

17
# import pdb
18
"""
cmaffeo2's avatar
cmaffeo2 committed
19
TODO:
cmaffeo2's avatar
cmaffeo2 committed
20
 + fix handling of crossovers for atomic representation
cmaffeo2's avatar
cmaffeo2 committed
21
 + map to atomic representation
22
    + add nicks
cmaffeo2's avatar
cmaffeo2 committed
23
    + transform ssDNA nucleotides 
cmaffeo2's avatar
cmaffeo2 committed
24
    - shrink ssDNA
cmaffeo2's avatar
cmaffeo2 committed
25
    + shrink dsDNA backbone
26
    + make orientation continuous
cmaffeo2's avatar
cmaffeo2 committed
27
    + sequence
28
    - handle circular dna
29
 + ensure crossover bead potentials aren't applied twice 
30
 + remove performance bottlenecks
31
 - test for large systems
cmaffeo2's avatar
cmaffeo2 committed
32
 + assign sequence
33
 + ENM
34
35
 - rework Location class 
 - remove recursive calls
36
 - document
37
 - add unit test of helices connected to themselves
38
39
"""

cmaffeo2's avatar
cmaffeo2 committed
40
41
42
class ParticleNotConnectedError(Exception):
    pass

43
44
class Location():
    """ Site for connection within an object """
45
    def __init__(self, container, address, type_, on_fwd_strand = True):
46
        ## TODO: remove cyclic references(?)
47
        self.container = container
cmaffeo2's avatar
cmaffeo2 committed
48
        self.address = address  # represents position along contour length in segment
cmaffeo2's avatar
cmaffeo2 committed
49
        # assert( type_ in ("end3","end5") ) # TODO remove or make conditional
50
        self.on_fwd_strand = on_fwd_strand
51
52
        self.type_ = type_
        self.particle = None
53
        self.connection = None
54
        self.is_3prime_side_of_connection = None
55

56
57
        self.prev_in_strand = None
        self.next_in_strand = None
58
59
        
        self.combine = None     # some locations might be combined in bead model 
60
61
62
63
64
65
66

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

67
    def set_connection(self, connection, is_3prime_side_of_connection):
68
        self.connection = connection # TODO weakref? 
69
        self.is_3prime_side_of_connection = is_3prime_side_of_connection
70

71
72
73
74
75
76
77
    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:
78
                pos = self.container.num_nt-1
79
80
81
82
            else:
                raise
        return pos

83
84
85
86
87
88
89
    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)
        
90
91
92
93
94
95
96
97
98
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_
        
99
100
101
102
103
104
105
    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
106

107
108
109
110
111
112
    def delete(self):
        self.A.container.connections.remove(self)
        self.B.container.connections.remove(self)
        self.A.connection = None
        self.B.connection = None

cmaffeo2's avatar
cmaffeo2 committed
113
114
115
116
    def __repr__(self):
        return "<Connection {}--{}--{}]>".format( self.A, self.type_, self.B )
        

117
        
118
119
120
# class ConnectableElement(Transformable):
class ConnectableElement():
    """ Abstract base class """
cmaffeo2's avatar
cmaffeo2 committed
121
122
123
124
    def __init__(self, connection_locations=None, connections=None):
        if connection_locations is None: connection_locations = []
        if connections is None: connections = []

125
126
        ## TODO decide on names
        self.locations = self.connection_locations = connection_locations
127
128
        self.connections = connections

cmaffeo2's avatar
cmaffeo2 committed
129
    def get_locations(self, type_=None, exclude=()):
130
131
132
133
134
135
136
137
        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] ) )
138
139
140
141
        return locs

    def get_location_at(self, address, on_fwd_strand=True, new_type="crossover"):
        loc = None
142
        if (self.num_nt == 1):
143
144
145
146
147
148
149
            # 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
150
            # assert( loc is not None )
151
152
153
154
155
156
157
158
        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
159

cmaffeo2's avatar
cmaffeo2 committed
160
    def get_connections_and_locations(self, connection_type=None, exclude=()):
161
162
        """ Returns a list with each entry of the form:
            connection, location_in_self, location_in_other """
163
        type_ = connection_type
164
165
        ret = []
        for c in self.connections:
166
            if (type_ is None or c.type_ == type_) and c.type_ not in exclude:
167
                if   c.A.container is self:
168
                    ret.append( [c, c.A, c.B] )
169
                elif c.B.container is self:
170
171
                    ret.append( [c, c.B, c.A] )
                else:
172
173
                    import pdb
                    pdb.set_trace()
174
175
176
                    raise Exception("Object contains connection that fails to refer to object")
        return ret

177
    def _connect(self, other, connection, in_3prime_direction=None):
178
179
        ## TODO fix circular references        
        A,B = [connection.A, connection.B]
180
181
182
183
        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
            
184
        A.connection = B.connection = connection
185
186
        self.connections.append(connection)
        other.connections.append(connection)
187
188
189
190
191
192
        l = A.container.locations
        if A not in l: l.append(A)
        l = B.container.locations
        if B not in l: l.append(B)
        

193
194
    # def _find_connections(self, loc):
    #     return [c for c in self.connections if c.A == loc or c.B == loc]
195
196
197

class SegmentParticle(PointParticle):
    def __init__(self, type_, position, name="A", segname="A", **kwargs):
198
        self.name = name
199
200
201
202
        self.contour_position = None
        PointParticle.__init__(self, type_, position, name=name, segname=segname, **kwargs)
        self.intrahelical_neighbors = []
        self.other_neighbors = []
cmaffeo2's avatar
cmaffeo2 committed
203
        self.locations = []
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218

    def get_intrahelical_above(self):
        """ Returns bead directly above self """
        assert( len(self.intrahelical_neighbors) <= 2 )
        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 """
        assert( len(self.intrahelical_neighbors) <= 2 )
        for b in self.intrahelical_neighbors:
            if b.get_contour_position(self.parent) < self.contour_position:
                return b

cmaffeo2's avatar
cmaffeo2 committed
219
220
221
222
223
    def _neighbor_should_be_added(self,b):
        c1 = self.contour_position
        c2 = b.get_contour_position(self.parent)
        if c2 < c1:
            b0 = self.get_intrahelical_below()
224
        else:
cmaffeo2's avatar
cmaffeo2 committed
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
            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:
            assert(len(b.intrahelical_neighbors) <= 1)
            assert(len(self.intrahelical_neighbors) <= 1)
            self.intrahelical_neighbors.append(b)
            b.intrahelical_neighbors.append(self)
246

cmaffeo2's avatar
cmaffeo2 committed
247
248
249
250
251
252
253
254
255
256
    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)

257
258
259
260
    def get_contour_position(self,seg):
        if seg == self.parent:
            return self.contour_position
        else:
cmaffeo2's avatar
cmaffeo2 committed
261
262
263
264
265
            cutoff = 30*3
            target_seg = seg

            ## depth-first search
            ## TODO cache distances to nearby locations?
cmaffeo2's avatar
cmaffeo2 committed
266
            def descend_search_tree(seg, contour_in_seg, distance=0, visited_segs=None):
cmaffeo2's avatar
cmaffeo2 committed
267
                nonlocal cutoff
cmaffeo2's avatar
cmaffeo2 committed
268
                if visited_segs is None: visited_segs = []
cmaffeo2's avatar
cmaffeo2 committed
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312

                if seg == target_seg:
                    # pdb.set_trace()
                    ## Found a segment in our target
                    sign = (contour_in_seg == 1) - (contour_in_seg == 0)
                    assert( sign in (1,-1) )
                    if distance < cutoff: # TODO: check if this does anything
                        cutoff = distance
                    return [[distance, contour_in_seg+sign*seg.nt_pos_to_contour(distance)]]
                if distance > cutoff:
                    return None
                    
                ret_list = []
                ## Find intrahelical locations in seg that we might pass through
                for c,A,B in seg.get_connections_and_locations("intrahelical"):
                    if B.container in visited_segs: continue
                    dx = seg.contour_to_nt_pos( np.abs(A.address-contour_in_seg),
                                                round_nt=False) 
                    results = descend_search_tree( B.container, B.address,
                                                   distance+dx, visited_segs + [seg] )
                    if results is not None:
                        ret_list.extend( results )
                return ret_list

            results = descend_search_tree(self.parent, self.contour_position)
            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)) )
            
313
314

## TODO break this class into smaller, better encapsulated pieces
315
316
317
318
319
320
321
322
323
324
325
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
326
327
328
329
330
    orientation_particle = ParticleType("O",
                                        diffusivity = 100,
                                        mass = 300,
                                        radius = 1,
                                    )
331

cmaffeo2's avatar
cmaffeo2 committed
332
    # orientation_bond = HarmonicBond(10,2)
333
    orientation_bond = HarmonicBond(30,1.5, rRange = (0,500) )
334
335
336
337
338
339
340

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

341
    def __init__(self, name, num_nt, 
cmaffeo2's avatar
cmaffeo2 committed
342
                 start_position = None,
343
344
345
                 end_position = None, 
                 segment_model = None):

cmaffeo2's avatar
cmaffeo2 committed
346
347
        if start_position is None: start_position = np.array((0,0,0))

348
        Group.__init__(self, name, children=[])
349
        ConnectableElement.__init__(self, connection_locations=[], connections=[])
350

351
        self.resname = name
cmaffeo2's avatar
cmaffeo2 committed
352
353
354
355
356
        self.start_orientation = None
        self.twist_per_nt = 0

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

357
358
359
        self._bead_model_generation = 0    # TODO: remove?
        self.segment_model = segment_model # TODO: remove?

360
361
362
363
        self.strand_pieces = dict()
        for d in ('fwd','rev'):
            self.strand_pieces[d] = []

364
        self.num_nt = int(num_nt)
365
        if end_position is None:
366
            end_position = np.array((0,0,self.distance_per_nt*num_nt)) + start_position
367
368
369
        self.start_position = start_position
        self.end_position = end_position

370
        ## Set up interpolation for positions
371
372
373
374
        self._set_splines_from_ends()

        self.sequence = None

375
376
377
    def __repr__(self):
        return "<{} {}[{:d}]>".format( type(self), self.name, self.num_nt )

378
379

    def _set_splines_from_ends(self):
380
381
382
        a = np.array([self.start_position,self.end_position]).T
        tck, u = interpolate.splprep( a, u=[0,1], s=0, k=1)
        self.position_spline_params = tck
383
        self.quaternion_spline_params = None
384

385

386
387
388
389
390
    def clear_all(self):
        Group.clear_all(self)  # TODO: use super?
        self.beads = []
        for c,loc,other in self.get_connections_and_locations():
            loc.particle = None
391

392
    def contour_to_nt_pos(self, contour_pos, round_nt=False):
393
        nt = contour_pos*(self.num_nt) - 0.5
394
        if round_nt:
cmaffeo2's avatar
cmaffeo2 committed
395
            assert( np.isclose(np.around(nt),nt) )
396
397
398
            nt = np.around(nt)
        return nt

399
    def nt_pos_to_contour(self,nt_pos):
400
        return (nt_pos+0.5)/(self.num_nt)
401

402
403
404
405
406
407
408
    def contour_to_position(self,s):
        p = interpolate.splev( s, self.position_spline_params )
        if len(p) > 1: p = np.array(p).T
        return p

    def contour_to_tangent(self,s):
        t = interpolate.splev( s, self.position_spline_params, der=1 )
409
410
        t = (t / np.linalg.norm(t,axis=0))
        return t.T
411
412
413
        

    def contour_to_orientation(self,s):
414
        assert( isinstance(s,float) or isinstance(s,int) or len(s) == 1 )   # TODO make vectorized version
415

416
417
418
419
420
        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)
421
            zAxis = np.array((0,0,1))
422
423

            if rotAxisL > 0.001:
424
425
426
                theta = np.arcsin(rotAxisL) * 180/np.pi
                if axis.dot(zAxis) < 0: theta = 180-theta
                orientation0 = rotationAboutAxis( rotAxis/rotAxisL, theta, normalizeAxis=False ).T
427
            else:
428
429
                orientation0 = np.eye(3) if axis.dot(zAxis) > 0 else \
                               rotationAboutAxis( np.array((1,0,0)), 180, normalizeAxis=False )
430
431
432
433
434
435
            orientation = rotationAboutAxis( axis, self.twist_per_nt*self.contour_to_nt_pos(s), normalizeAxis=False )
            orientation = orientation.dot(orientation0)
        else:
            q = interpolate.splev( s, self.quaternion_spline_params )
            if len(q) > 1: q = np.array(q).T # TODO: is this needed?
            orientation = quaternion_to_matrix(q)
436

437
        return orientation
438

cmaffeo2's avatar
cmaffeo2 committed
439
    def get_contour_sorted_connections_and_locations(self,type_):
cmaffeo2's avatar
cmaffeo2 committed
440
        sort_fn = lambda c: c[1].address
cmaffeo2's avatar
cmaffeo2 committed
441
        cl = self.get_connections_and_locations(type_)
cmaffeo2's avatar
cmaffeo2 committed
442
        return sorted(cl, key=sort_fn)
443
444
445
    
    def randomize_unset_sequence(self):
        bases = list(seqComplement.keys())
446
        # bases = ['T']        ## FOR DEBUG
447
        if self.sequence is None:
448
            self.sequence = [random.choice(bases) for i in range(self.num_nt)]
449
        else:
450
            assert(len(self.sequence) == self.num_nt) # TODO move
451
452
453
            for i in range(len(self.sequence)):
                if self.sequence[i] is None:
                    self.sequence[i] = random.choice(bases)
454

cmaffeo2's avatar
cmaffeo2 committed
455
456
457
    def _get_num_beads(self, max_basepairs_per_bead, max_nucleotides_per_bead ):
        raise NotImplementedError

458
    def _generate_one_bead(self, contour_position, nts):
459
460
        raise NotImplementedError

461
    def _generate_atomic_nucleotide(self, contour_position, is_fwd, seq, scale):
cmaffeo2's avatar
cmaffeo2 committed
462
463
464
465
466
        """ 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)
467
        orientation = self.contour_to_orientation(contour_position)
cmaffeo2's avatar
cmaffeo2 committed
468

469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
        """ 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
487
        key = seq
488
489
        nt_dict = canonicalNtFwd if is_fwd else canonicalNtRev

490
        atoms = nt_dict[ key ].generate() # TODO: clone?
cmaffeo2's avatar
cmaffeo2 committed
491
                        
cmaffeo2's avatar
cmaffeo2 committed
492
        atoms.orientation = orientation.dot(atoms.orientation)
493
494
495
496
497
        if isinstance(self, SingleStrandedSegment):
            if scale is not None and scale != 1:
                for a in atoms:
                    a.position = scale*a.position
                    a.beta = 0
498
            atoms.position = pos - atoms.atoms_by_name["C1'"].collapsedPosition()
499
500
501
502
503
504
505
506
507
508
509
        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
                        a.beta = 0
            atoms.position = pos
cmaffeo2's avatar
cmaffeo2 committed
510
511

        return atoms
512

513
514
    def add_location(self, nt, type_, on_fwd_strand=True):
        ## Create location if needed, add to segment
515
        c = self.nt_pos_to_contour(nt)
516
517
518
519
520
521
522
        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?
    def add_5prime(self, nt, on_fwd_strand=True):
523
524
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
525
        self.add_location(nt,"5prime",on_fwd_strand)
526
527

    def add_3prime(self, nt, on_fwd_strand=True):
528
529
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
530
        self.add_location(nt,"3prime",on_fwd_strand)
531

532
    def get_3prime_locations(self):
cmaffeo2's avatar
cmaffeo2 committed
533
        return sorted(self.get_locations("3prime"),key=lambda x: x.address)
534
    
cmaffeo2's avatar
cmaffeo2 committed
535
    def get_5prime_locations(self):
536
        ## TODO? ensure that data is consistent before _build_model calls
cmaffeo2's avatar
cmaffeo2 committed
537
        return sorted(self.get_locations("5prime"),key=lambda x: x.address)
cmaffeo2's avatar
cmaffeo2 committed
538

539
    def iterate_connections_and_locations(self, reverse=False):
cmaffeo2's avatar
cmaffeo2 committed
540
541
        ## connections to other segments
        cl = self.get_contour_sorted_connections_and_locations()
542
        if reverse:
cmaffeo2's avatar
cmaffeo2 committed
543
            cl = cl[::-1]
544
545
546
            
        for c in cl:
            yield c
cmaffeo2's avatar
cmaffeo2 committed
547

548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
    ## 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)

566
    ## TODO rename
567
    def get_strand_segment(self, nt_pos, is_fwd, move_at_least=0.5):
568
        """ Walks through locations, checking for crossovers """
569
570
571
572
        # if self.name in ("6-1","1-1"):
        #     import pdb
        #     pdb.set_trace()
        move_at_least = 0
573
574

        ## Iterate through locations
cmaffeo2's avatar
cmaffeo2 committed
575
        # locations = sorted(self.locations, key=lambda l:(l.address,not l.on_fwd_strand), reverse=(not is_fwd))
576
577
578
579
580
581
        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))
582
583
        # print(locations)

584
        for l in locations:
cmaffeo2's avatar
cmaffeo2 committed
585
586
587
588
            # TODOTODO probably okay
            if l.address == 0:
                pos = 0.0
            elif l.address == 1:
589
                pos = self.num_nt-1
cmaffeo2's avatar
cmaffeo2 committed
590
591
            else:
                pos = self.contour_to_nt_pos(l.address, round_nt=True)
592
593
594

            ## DEBUG

cmaffeo2's avatar
cmaffeo2 committed
595

596
            ## Skip locations encountered before our strand
597
598
599
600
601
602
603
604
            # 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
605
606
607

            ## Stop if we found the 3prime end
            if l.on_fwd_strand == is_fwd and l.type_ == "3prime":
608
                # print("  found end at",l)
609
                return pos, None, None, None, None
610
611
612
613
614
615
616
617

            ## 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:
618
619
                # print("  passing through",l)
                # print("from {}, connection {} to {}".format(nt_pos,l,B))
620
                Bpos = B.get_nt_pos()
621
                return pos, B.container, Bpos, B.on_fwd_strand, 0.5
622
623
624
                
            ## Stop at other strand crossovers so basepairs line up
            elif c.type_ == "crossover":
625
                if nt_pos == pos: continue
626
                # print("  pausing at",l)
627
                return pos, l.container, pos+(2*is_fwd-1), is_fwd, 0
628
629
630
631
632
633

        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

634
635
636
    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
637
        # TODO: include beads in connections?
638
639
640
        i = np.argmin((cs - contour_position)**2)

        return self.beads[i]
641

642
643
644
645
646
647
648
649
650
    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 ))

651
652
    def get_all_consecutive_beads(self, number):
        assert(number >= 1)
cmaffeo2's avatar
cmaffeo2 committed
653
        ## Assume that consecutive beads in self.beads are bonded
654
        ret = []
cmaffeo2's avatar
cmaffeo2 committed
655
656
        for i in range(len(self.beads)-number+1):
            tmp = [self.beads[i+j] for j in range(0,number)]
657
            ret.append( tmp )
658
        return ret   
659

660
661
662
    def _add_bead(self,b,set_contour=False):
        if set_contour:
            b.contour_position = b.get_contour_position(self)
663
        
664
665
666
        # assert(b.parent is None)
        if b.parent is not None:
            b.parent.children.remove(b)
667
        self.add(b)
668
669
670
671
672
673
        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)
674
            self.add(o)
675
676
677
678
679
680
681
682
            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 = []
683
684

        if True:
685
686
            ## TODO: remove this if duplicates are never found 
            # print("Searching for duplicate particles...")
687
            ## Remove duplicates, preserving order
688
689
690
691
            tmp = []
            for c in new_children:
                if c not in tmp:
                    tmp.append(c)
692
                else:
693
                    print("  DUPLICATE PARTICLE FOUND!")
694
695
            new_children = tmp

696
697
698
699
700
        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)
701
702
703
704
705
            
        # 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)
706
707
        assert(len(old_children) == len(self.children))
        assert(len(old_beads) == len(self.beads))
708

709

cmaffeo2's avatar
cmaffeo2 committed
710
    def _generate_beads(self, bead_model, max_basepairs_per_bead, max_nucleotides_per_bead):
711

712
        """ Generate beads (positions, types, etc) and bonds, angles, dihedrals, exclusions """
cmaffeo2's avatar
cmaffeo2 committed
713
        ## TODO: decide whether to remove bead_model argument
714
        ##       (currently unused)
cmaffeo2's avatar
cmaffeo2 committed
715

716
        ## First find points between-which beads must be generated
717
718
719
        # 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
720
721
        # if self.name == "31-2":
            # pdb.set_trace()
722

723
724
725
726
727
        existing_beads = {l.particle for l in self.locations if l.particle is not None}
        existing_beads = sorted( list(existing_beads), key=lambda b: b.get_contour_position(self) )
        
        if len(existing_beads) != len(set(existing_beads)):
            pdb.set_trace()
728
729
730
731
        for b in existing_beads:
            assert(b.parent is not None)

        ## Add ends if they don't exist yet
732
        ## TODOTODO: test 1 nt segments?
733
        if len(existing_beads) == 0 or existing_beads[0].get_nt_position(self) >= 0.5:
cmaffeo2's avatar
cmaffeo2 committed
734
735
            # if len(existing_beads) > 0:            
            #     assert(existing_beads[0].get_nt_position(self) >= 0.5)
736
            b = self._generate_one_bead( self.nt_pos_to_contour(0), 0)
737
            existing_beads = [b] + existing_beads
cmaffeo2's avatar
cmaffeo2 committed
738

739
740
        if existing_beads[-1].get_nt_position(self)-(self.num_nt-1) < -0.5:
            b = self._generate_one_bead( self.nt_pos_to_contour(self.num_nt-1), 0)
741
742
743
744
745
            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
746
        last = None
cmaffeo2's avatar
cmaffeo2 committed
747

748
749
        for I in range(len(existing_beads)-1):
            eb1,eb2 = [existing_beads[i] for i in (I,I+1)]
750
            assert( eb1 is not eb2 )
751

cmaffeo2's avatar
cmaffeo2 committed
752
753
754
755
            # if np.isclose(eb1.position[2], eb2.position[2]):
            #     import pdb
            #     pdb.set_trace()

756
            # print(" %s working on %d to %d" % (self.name, eb1.position[2], eb2.position[2]))
757
758
            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
759
760
761
762
763
764

            ## 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

765
            ds = e_ds / (num_beads+1)
766
767
768
            nts = ds*self.num_nt
            eb1.num_nt += 0.5*nts
            eb2.num_nt += 0.5*nts
769
770
771
772
773
774
775

            ## 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
776
                last.make_intrahelical_neighbor(eb1)
777
778
779
            last = eb1
            for j in range(num_beads):
                s = ds*(j+1) + s0
780
                # if self.name in ("51-2","51-3"):
cmaffeo2's avatar
cmaffeo2 committed
781
                # if self.name in ("31-2",):
782
                #     print(" adding bead at {}".format(s))
783
784
                b = self._generate_one_bead(s,nts)

cmaffeo2's avatar
cmaffeo2 committed
785
                last.make_intrahelical_neighbor(b)
786
787
788
                last = b
                tmp_children.append(b)

cmaffeo2's avatar
cmaffeo2 committed
789
        last.make_intrahelical_neighbor(eb2)
790
791
792

        if eb2.parent == self:
            tmp_children.append(eb2)
cmaffeo2's avatar
cmaffeo2 committed
793
        # if self.name in ("31-2",):
794
        #     pdb.set_trace()
795
        self._rebuild_children(tmp_children)
796
797
798
799
800
801
802
803
804
805

    def _regenerate_beads(self, max_nts_per_bead=4, ):
        ...
    

class DoubleStrandedSegment(Segment):

    """ Class that describes a segment of ssDNA. When built from
    cadnano models, should not span helices """

806
    def __init__(self, name, num_bp, start_position = np.array((0,0,0)),
807
808
                 end_position = None, 
                 segment_model = None,
cmaffeo2's avatar
cmaffeo2 committed
809
810
                 local_twist = False,
                 num_turns = None,
cmaffeo2's avatar
cmaffeo2 committed
811
812
                 start_orientation = None,
                 twist_persistence_length = 90 ):
cmaffeo2's avatar
cmaffeo2 committed
813
814
815
        
        self.helical_rise = 10.44
        self.distance_per_nt = 3.4
816
        Segment.__init__(self, name, num_bp,
817
818
819
                         start_position,
                         end_position, 
                         segment_model)
820
        self.num_bp = self.num_nt
821

cmaffeo2's avatar
cmaffeo2 committed
822
823
        self.local_twist = local_twist
        if num_turns is None:
824
825
            num_turns = float(num_bp) / self.helical_rise
        self.twist_per_nt = float(360 * num_turns) / num_bp
cmaffeo2's avatar
cmaffeo2 committed
826
827

        if start_orientation is None:
828
            start_orientation = np.eye(3) # np.array(((1,0,0),(0,1,0),(0,0,1)))
cmaffeo2's avatar
cmaffeo2 committed
829
        self.start_orientation = start_orientation
cmaffeo2's avatar
cmaffeo2 committed
830
        self.twist_persistence_length = twist_persistence_length
cmaffeo2's avatar
cmaffeo2 committed
831

832
833
        self.nicks = []

834
        self.start = self.start5 = Location( self, address=0, type_= "end5" )
835
        self.start3 = Location( self, address=0, type_ = "end3", on_fwd_strand=False )
836

837
838
        self.end = self.end3 = Location( self, address=1, type_ = "end3" )
        self.end5 = Location( self, address=1, type_= "end5", on_fwd_strand=False )
cmaffeo2's avatar
cmaffeo2 committed
839
840
        # for l in (self.start5,self.start3,self.end3,self.end5):
        #     self.locations.append(l)
841

842
843
844
845
846
847
848
        ## Set up interpolation for azimuthal angles 
        a = np.array([self.start_position,self.end_position]).T
        tck, u = interpolate.splprep( a, u=[0,1], s=0, k=1)
        self.position_spline_params = tck
        
        ## TODO: initialize sensible spline for orientation

849
    ## Convenience methods
850
    ## TODO: add errors if unrealistic connections are made
851
    ## TODO: make connections automatically between unconnected strands
852
    def connect_start5(self, end3, type_="intrahelical", force_connection=False):
853
854
        if isinstance(end3, SingleStrandedSegment):
            end3 = end3.end3
855
856
        self._connect_ends( self.start5, end3, type_, force_connection = force_connection )
    def connect_start3(self, end5, type_="intrahelical", force_connection=False):
857
        if isinstance(end5, SingleStrandedSegment):
858
            end5 = end5.start5
859
860
        self._connect_ends( self.start3, end5, type_, force_connection = force_connection )
    def connect_end3(self, end5, type_="intrahelical", force_connection=False):
861
        if isinstance(end5, SingleStrandedSegment):
862
            end5 = end5.start5
863
864
        self._connect_ends( self.end3, end5, type_, force_connection = force_connection )
    def connect_end5(self, end3, type_="intrahelical", force_connection=False):
865
866
        if isinstance(end3, SingleStrandedSegment):
            end3 = end3.end3
867
        self._connect_ends( self.end5, end3, type_, force_connection = force_connection )
868

869
    def add_crossover(self, nt, other, other_nt, strands_fwd=(True,False), nt_on_5prime=True, type_="crossover"):
cmaffeo2's avatar
cmaffeo2 committed
870
871
872
873
        """ Add a crossover between two helices """
        ## Validate other, nt, other_nt
        ##   TODO

874
        if isinstance(other,SingleStrandedSegment):
cmaffeo2's avatar
cmaffeo2 committed
875
            other.add_crossover(other_nt, self, nt, strands_fwd[::-1], not nt_on_5prime)
876
        else:
877

878
879
880
            ## Create locations, connections and add to segments
            c = self.nt_pos_to_contour(nt)
            assert(c >= 0 and c <= 1)
881

882
883
884
            loc = self.get_location_at(c, strands_fwd[0])

            c = other.nt_pos_to_contour(other_nt)
cmaffeo2's avatar
cmaffeo2 committed
885
            # TODOTODO: may need to subtract or add a little depending on 3prime/5prime
886
887
            assert(c >= 0 and c <= 1)
            other_loc = other.get_location_at(c, strands_fwd[1])
888
            self._connect(other, Connection( loc, other_loc, type_=type_ ))
cmaffeo2's avatar
cmaffeo2 committed
889
890
891
892
893
894
            if nt_on_5prime:
                loc.is_3prime_side_of_connection = False
                other_loc.is_3prime_side_of_connection = True
            else:            
                loc.is_3prime_side_of_connection = True
                other_loc.is_3prime_side_of_connection = False
cmaffeo2's avatar
cmaffeo2 committed
895

896
    ## Real work
897
    def _connect_ends(self, end1, end2, type_, force_connection):
898
        ## TODO remove self?
899
900
901
902
903
        ## validate the input
        for end in (end1, end2):
            assert( isinstance(end, Location) )
            assert( end.type_ in ("end3","end5") )
        assert( end1.type_ != end2.type_ )
904
905
906
907
908
909
910
911
912

        ## Remove other connections involving these points
        if end1.connection is not None:
            print("WARNING: reconnecting {}".format(end1))
            end1.connection.delete()
        if end2.connection is not None:
            print("WARNING: reconnecting {}".format(end2))
            end2.connection.delete()

913
        ## Create and add connection
cmaffeo2's avatar
cmaffeo2 committed
914
        if end2.type_ == "end5":
915
916
917
            end1.container._connect( end2.container, Connection( end1, end2, type_=type_ ), in_3prime_direction=True )
        else:
            end2.container._connect( end1.container, Connection( end2, end1, type_=type_ ), in_3prime_direction=True )
918
    def _get_num_beads(self, contour, max_basepairs_per_bead, max_nucleotides_per_bead):
919
920
        # return int(contour*self.num_nt // max_basepairs_per_bead)
        return int(contour*(self.num_nt**2/(self.num_nt+1)) // max_basepairs_per_bead)
cmaffeo2's avatar
cmaffeo2 committed
921

922
923
    def _generate_one_bead(self, contour_position, nts):
        pos = self.contour_to_position(contour_position)
cmaffeo2's avatar
cmaffeo2 committed
924
        if self.local_twist:
925
            orientation = self.contour_to_orientation(contour_position)
cmaffeo2's avatar
cmaffeo2 committed
926
927
928
            if orientation is None:
                print("WARNING: local_twist is True, but orientation is None; using identity")
                orientation = np.eye(3)
cmaffeo2's avatar
cmaffeo2 committed
929
            opos = pos + orientation.dot( np.array((Segment.orientation_bond.r0,0,0)) )
cmaffeo2's avatar
cmaffeo2 committed
930
931
932
933
934
            # if np.linalg.norm(pos) > 1e3:
            #     pdb.set_trace()
            assert(np.linalg.norm(opos-pos) < 10 )
            o = SegmentParticle( Segment.orientation_particle, opos, name="O",
                                 contour_position = contour_position,
935
                                 num_nt=nts, parent=self )
936
            bead = SegmentParticle( Segment.dsDNA_particle, pos, name="DNA",
937
                                    num_nt=nts, parent=self, 
938
939
                                    orientation_bead=o,
                                    contour_position=contour_position )
cmaffeo2's avatar
cmaffeo2 committed
940
941

        else:
942
            bead = SegmentParticle( Segment.dsDNA_particle, pos, name="DNA",
943
                                    num_nt=nts, parent=self,
944
945
                                    contour_position=contour_position )
        self._add_bead(bead)
cmaffeo2's avatar
cmaffeo2 committed
946
        return bead
947
948
949
950
951
952

class SingleStrandedSegment(Segment):

    """ Class that describes a segment of ssDNA. When built from
    cadnano models, should not span helices """

953
    def __init__(self, name, num_nt, start_position = None,
954
955
956
                 end_position = None, 
                 segment_model = None):

cmaffeo2's avatar
cmaffeo2 committed
957
        if start_position is None: start_position = np.array((0,0,0))
958
        self.distance_per_nt = 5
959
        Segment.__init__(self, name, num_nt, 
960
961
962
963
                         start_position,
                         end_position, 
                         segment_model)

964
        self.start = self.start5 = Location( self, address=0, type_= "end5" ) # TODO change type_?
965
        self.end = self.end3 = Location( self, address=1, type_ = "end3" )
cmaffeo2's avatar
cmaffeo2 committed
966
967
        # for l in (self.start5,self.end3):
        #     self.locations.append(l)
968

969
    def connect_end3(self, end5, force_connection=False):
cmaffeo2's avatar
cmaffeo2 committed
970
        self._connect_end( end5,  _5_to_3 = True, force_connection = force_connection )
971

972
    def connect_5end(self, end3, force_connection=False): # TODO: change name or possibly deprecate
cmaffeo2's avatar
cmaffeo2 committed
973
        self._connect_end( end3,  _5_to_3 = False, force_connection = force_connection )
974
975
976
977

    def _connect_end(self, other, _5_to_3, force_connection):
        assert( isinstance(other, Location) )
        if _5_to_3 == True:
978
979
980
981
            seg1 = self
            seg2 = other.container
            end1 = self.end3
            end2 = other