segmentmodel.py 94.6 KB
Newer Older
cmaffeo2's avatar
cmaffeo2 committed
1
import pdb
2
import numpy as np
3
import random
4
from arbdmodel import PointParticle, ParticleType, Group, ArbdModel
5
from coords import rotationAboutAxis, quaternion_from_matrix, quaternion_to_matrix
6
7
8
9
from nonbonded import *
from copy import copy, deepcopy
from nbPot import nbDnaScheme

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

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

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
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:
                pos = self.container.num_nts-1
            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

    def __repr__(self):
        return "<Connection {}--{}--{}]>".format( self.A, self.type_, self.B )
        

111
        
112
113
114
# class ConnectableElement(Transformable):
class ConnectableElement():
    """ Abstract base class """
cmaffeo2's avatar
cmaffeo2 committed
115
116
117
118
    def __init__(self, connection_locations=None, connections=None):
        if connection_locations is None: connection_locations = []
        if connections is None: connections = []

119
120
        ## TODO decide on names
        self.locations = self.connection_locations = connection_locations
121
122
        self.connections = connections

cmaffeo2's avatar
cmaffeo2 committed
123
    def get_locations(self, type_=None, exclude=()):
124
125
126
127
128
129
130
131
        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] ) )
132
133
134
135
136
137
138
139
140
141
142
143
        return locs

    def get_location_at(self, address, on_fwd_strand=True, new_type="crossover"):
        loc = None
        if (self.num_nts == 1):
            # 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
144
            # assert( loc is not None )
145
146
147
148
149
150
151
152
        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
153

cmaffeo2's avatar
cmaffeo2 committed
154
    def get_connections_and_locations(self, connection_type=None, exclude=()):
155
156
        """ Returns a list with each entry of the form:
            connection, location_in_self, location_in_other """
157
        type_ = connection_type
158
159
        ret = []
        for c in self.connections:
160
            if (type_ is None or c.type_ == type_) and c.type_ not in exclude:
161
                if   c.A.container is self:
162
                    ret.append( [c, c.A, c.B] )
163
                elif c.B.container is self:
164
165
                    ret.append( [c, c.B, c.A] )
                else:
166
167
                    import pdb
                    pdb.set_trace()
168
169
170
                    raise Exception("Object contains connection that fails to refer to object")
        return ret

171
    def _connect(self, other, connection, in_3prime_direction=None):
172
173
        ## TODO fix circular references        
        A,B = [connection.A, connection.B]
174
175
176
177
        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
            
178
        A.connection = B.connection = connection
179
180
        self.connections.append(connection)
        other.connections.append(connection)
181
182
183
184
185
186
        l = A.container.locations
        if A not in l: l.append(A)
        l = B.container.locations
        if B not in l: l.append(B)
        

187
188
    # def _find_connections(self, loc):
    #     return [c for c in self.connections if c.A == loc or c.B == loc]
189
190
191

class SegmentParticle(PointParticle):
    def __init__(self, type_, position, name="A", segname="A", **kwargs):
192
        self.name = name
193
194
195
196
        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
197
        self.locations = []
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212

    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
213
214
215
216
217
    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()
218
        else:
cmaffeo2's avatar
cmaffeo2 committed
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
            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)
240

cmaffeo2's avatar
cmaffeo2 committed
241
242
243
244
245
246
247
248
249
250
    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)

251
252
253
254
    def get_contour_position(self,seg):
        if seg == self.parent:
            return self.contour_position
        else:
cmaffeo2's avatar
cmaffeo2 committed
255
256
257
258
259
            cutoff = 30*3
            target_seg = seg

            ## depth-first search
            ## TODO cache distances to nearby locations?
cmaffeo2's avatar
cmaffeo2 committed
260
            def descend_search_tree(seg, contour_in_seg, distance=0, visited_segs=None):
cmaffeo2's avatar
cmaffeo2 committed
261
                nonlocal cutoff
cmaffeo2's avatar
cmaffeo2 committed
262
                if visited_segs is None: visited_segs = []
cmaffeo2's avatar
cmaffeo2 committed
263
264
265
266
267
268
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

                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)) )
            
307
308

## TODO break this class into smaller, better encapsulated pieces
309
310
311
312
313
314
315
316
317
318
319
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
320
321
322
323
324
    orientation_particle = ParticleType("O",
                                        diffusivity = 100,
                                        mass = 300,
                                        radius = 1,
                                    )
325

cmaffeo2's avatar
cmaffeo2 committed
326
    # orientation_bond = HarmonicBond(10,2)
327
    orientation_bond = HarmonicBond(30,1.5, rRange = (0,500) )
328
329
330
331
332
333
334
335

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

    def __init__(self, name, num_nts, 
cmaffeo2's avatar
cmaffeo2 committed
336
                 start_position = None,
337
338
339
                 end_position = None, 
                 segment_model = None):

cmaffeo2's avatar
cmaffeo2 committed
340
341
        if start_position is None: start_position = np.array((0,0,0))

342
        Group.__init__(self, name, children=[])
343
        ConnectableElement.__init__(self, connection_locations=[], connections=[])
344

345
        self.resname = name
cmaffeo2's avatar
cmaffeo2 committed
346
347
348
349
350
        self.start_orientation = None
        self.twist_per_nt = 0

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

351
352
353
        self._bead_model_generation = 0    # TODO: remove?
        self.segment_model = segment_model # TODO: remove?

cmaffeo2's avatar
cmaffeo2 committed
354
        self.num_nts = int(num_nts)
355
356
357
358
359
        if end_position is None:
            end_position = np.array((0,0,self.distance_per_nt*num_nts)) + start_position
        self.start_position = start_position
        self.end_position = end_position

360
361
362
363
        ## Set up interpolation for positions
        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
364
        self.quaternion_spline_params = None
365
366
        
        self.sequence = None
367

368

369
370
371
372
373
    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
374

375
    def contour_to_nt_pos(self, contour_pos, round_nt=False):
cmaffeo2's avatar
cmaffeo2 committed
376
        nt = contour_pos*(self.num_nts) - 0.5
377
        if round_nt:
cmaffeo2's avatar
cmaffeo2 committed
378
            assert( np.isclose(np.around(nt),nt) )
379
380
381
            nt = np.around(nt)
        return nt

382
    def nt_pos_to_contour(self,nt_pos):
cmaffeo2's avatar
cmaffeo2 committed
383
        return (nt_pos+0.5)/(self.num_nts)
384

385
386
387
388
389
390
391
    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 )
392
393
        t = (t / np.linalg.norm(t,axis=0))
        return t.T
394
395
396
        

    def contour_to_orientation(self,s):
397
398
        assert( isinstance(s,float) or isinstance(s,int) or len(s) == 1 )   # TODO make vectorized version
        orientation = None
399
400
401
402
403
404
405
406
407
        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)
            theta = np.arcsin(rotAxisL)

            if rotAxisL > 0.001:
                orientation0 = rotationAboutAxis( rotAxis/rotAxisL, theta*180/np.pi, normalizeAxis=False )
408
            else:
409
410
411
412
413
414
415
                orientation0 = np.eye(3)
            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)
416
        return orientation
417

cmaffeo2's avatar
cmaffeo2 committed
418
    def get_contour_sorted_connections_and_locations(self,type_):
cmaffeo2's avatar
cmaffeo2 committed
419
        sort_fn = lambda c: c[1].address
cmaffeo2's avatar
cmaffeo2 committed
420
        cl = self.get_connections_and_locations(type_)
cmaffeo2's avatar
cmaffeo2 committed
421
        return sorted(cl, key=sort_fn)
422
423
424
    
    def randomize_unset_sequence(self):
        bases = list(seqComplement.keys())
425
        # bases = ['T']        ## FOR DEBUG
426
427
428
429
430
431
432
        if self.sequence is None:
            self.sequence = [random.choice(bases) for i in range(self.num_nts)]
        else:
            assert(len(self.sequence) == self.num_nts) # TODO move
            for i in range(len(self.sequence)):
                if self.sequence[i] is None:
                    self.sequence[i] = random.choice(bases)
433

cmaffeo2's avatar
cmaffeo2 committed
434
435
436
    def _get_num_beads(self, max_basepairs_per_bead, max_nucleotides_per_bead ):
        raise NotImplementedError

437
    def _generate_one_bead(self, contour_position, nts):
438
439
        raise NotImplementedError

440
    def _generate_atomic_nucleotide(self, contour_position, is_fwd, seq, scale):
cmaffeo2's avatar
cmaffeo2 committed
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
        """ 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)
        if self.local_twist:
            orientation = self.contour_to_orientation(contour_position)
            ## TODO: move this code (?)
            if orientation is None:
                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
457
                ## TODO: improve placement of ssDNA
cmaffeo2's avatar
cmaffeo2 committed
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
                # rot = rotationAboutAxis( axis, contour_position*self.twist_per_nt*self.num_nts, normalizeAxis=True )
                # orientation = rot.dot(orientation)
            else:
                orientation = orientation
                            
        else:
            raise NotImplementedError

        # key = self.sequence
        # if self.ntAt5prime is None and self.ntAt3prime is not None: key = "5"+key
        # if self.ntAt5prime is not None and self.ntAt3prime is None: key = key+"3"
        # if self.ntAt5prime is None and self.ntAt3prime is None: key = key+"singlet"

        key = seq
        if not is_fwd:
            nt_dict = canonicalNtFwd
        else:
            nt_dict = canonicalNtRev
476
        atoms = nt_dict[ key ].generate() # TODO: clone?
cmaffeo2's avatar
cmaffeo2 committed
477
                        
cmaffeo2's avatar
cmaffeo2 committed
478
        atoms.orientation = orientation.dot(atoms.orientation)
479
480
481
482
483
        if isinstance(self, SingleStrandedSegment):
            if scale is not None and scale != 1:
                for a in atoms:
                    a.position = scale*a.position
                    a.beta = 0
484
            atoms.position = pos - atoms.atoms_by_name["C1'"].collapsedPosition()
485
486
487
488
489
490
491
492
493
494
495
        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
496
497

        return atoms
498

499
500
    def add_location(self, nt, type_, on_fwd_strand=True):
        ## Create location if needed, add to segment
501
        c = self.nt_pos_to_contour(nt)
502
503
504
505
506
507
508
        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):
509
510
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
511
        self.add_location(nt,"5prime",on_fwd_strand)
512
513

    def add_3prime(self, nt, on_fwd_strand=True):
514
515
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
516
        self.add_location(nt,"3prime",on_fwd_strand)
517

518
    def get_3prime_locations(self):
cmaffeo2's avatar
cmaffeo2 committed
519
        return sorted(self.get_locations("3prime"),key=lambda x: x.address)
520
    
cmaffeo2's avatar
cmaffeo2 committed
521
    def get_5prime_locations(self):
522
        ## TODO? ensure that data is consistent before _build_model calls
cmaffeo2's avatar
cmaffeo2 committed
523
        return sorted(self.get_locations("5prime"),key=lambda x: x.address)
cmaffeo2's avatar
cmaffeo2 committed
524

525
    def iterate_connections_and_locations(self, reverse=False):
cmaffeo2's avatar
cmaffeo2 committed
526
527
        ## connections to other segments
        cl = self.get_contour_sorted_connections_and_locations()
528
        if reverse:
cmaffeo2's avatar
cmaffeo2 committed
529
            cl = cl[::-1]
530
531
532
            
        for c in cl:
            yield c
cmaffeo2's avatar
cmaffeo2 committed
533

534
    ## TODO rename
535
    def get_strand_segment(self, nt_pos, is_fwd, move_at_least=0.5):
536
        """ Walks through locations, checking for crossovers """
537
538
539
540
        # if self.name in ("6-1","1-1"):
        #     import pdb
        #     pdb.set_trace()
        move_at_least = 0
541
542

        ## Iterate through locations
cmaffeo2's avatar
cmaffeo2 committed
543
        # locations = sorted(self.locations, key=lambda l:(l.address,not l.on_fwd_strand), reverse=(not is_fwd))
544
545
546
547
548
549
        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))
550
551
        # print(locations)

552
        for l in locations:
cmaffeo2's avatar
cmaffeo2 committed
553
554
555
556
557
558
559
            # TODOTODO probably okay
            if l.address == 0:
                pos = 0.0
            elif l.address == 1:
                pos = self.num_nts-1
            else:
                pos = self.contour_to_nt_pos(l.address, round_nt=True)
560
561
562

            ## DEBUG

cmaffeo2's avatar
cmaffeo2 committed
563

564
            ## Skip locations encountered before our strand
565
566
567
568
569
570
571
572
            # 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
573
574
575

            ## Stop if we found the 3prime end
            if l.on_fwd_strand == is_fwd and l.type_ == "3prime":
576
                # print("  found end at",l)
577
                return pos, None, None, None, None
578
579
580
581
582
583
584
585

            ## 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:
586
587
                # print("  passing through",l)
                # print("from {}, connection {} to {}".format(nt_pos,l,B))
588
                Bpos = B.get_nt_pos()
589
                return pos, B.container, Bpos, B.on_fwd_strand, 0.5
590
591
592
                
            ## Stop at other strand crossovers so basepairs line up
            elif c.type_ == "crossover":
593
                if nt_pos == pos: continue
594
                # print("  pausing at",l)
595
                return pos, l.container, pos+(2*is_fwd-1), is_fwd, 0
596

597
598
        import pdb
        pdb.set_trace()
599
600
601
602
603
        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

604
605
606
    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
607
        # TODO: include beads in connections?
608
609
610
        i = np.argmin((cs - contour_position)**2)

        return self.beads[i]
611
612
613

    def get_all_consecutive_beads(self, number):
        assert(number >= 1)
cmaffeo2's avatar
cmaffeo2 committed
614
        ## Assume that consecutive beads in self.beads are bonded
615
        ret = []
cmaffeo2's avatar
cmaffeo2 committed
616
617
        for i in range(len(self.beads)-number+1):
            tmp = [self.beads[i+j] for j in range(0,number)]
618
            ret.append( tmp )
619
        return ret   
620

621
622
623
    def _add_bead(self,b,set_contour=False):
        if set_contour:
            b.contour_position = b.get_contour_position(self)
624
        
625
626
627
        # assert(b.parent is None)
        if b.parent is not None:
            b.parent.children.remove(b)
628
        self.add(b)
629
630
631
632
633
634
        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)
635
            self.add(o)
636
637
638
639
640
641
642
643
            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 = []
644
645

        if True:
646
647
            ## TODO: remove this if duplicates are never found 
            # print("Searching for duplicate particles...")
648
            ## Remove duplicates, preserving order
649
650
651
652
            tmp = []
            for c in new_children:
                if c not in tmp:
                    tmp.append(c)
653
                else:
654
                    print("  DUPLICATE PARTICLE FOUND!")
655
656
            new_children = tmp

657
658
659
660
661
        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)
662
663
664
665
666
            
        # 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)
667
668
        assert(len(old_children) == len(self.children))
        assert(len(old_beads) == len(self.beads))
669

670

cmaffeo2's avatar
cmaffeo2 committed
671
    def _generate_beads(self, bead_model, max_basepairs_per_bead, max_nucleotides_per_bead):
672

673
        """ Generate beads (positions, types, etc) and bonds, angles, dihedrals, exclusions """
cmaffeo2's avatar
cmaffeo2 committed
674
        ## TODO: decide whether to remove bead_model argument
675
        ##       (currently unused)
cmaffeo2's avatar
cmaffeo2 committed
676

677
        ## First find points between-which beads must be generated
678
679
680
        # 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
681
682
        # if self.name == "31-2":
            # pdb.set_trace()
683

684
685
686
687
688
        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()
689
690
691
        for b in existing_beads:
            assert(b.parent is not None)

cmaffeo2's avatar
cmaffeo2 committed
692
        # if self.name in ("31-2",):
cmaffeo2's avatar
cmaffeo2 committed
693
694
        #     pdb.set_trace()

695
        ## Add ends if they don't exist yet
696
        ## TODOTODO: test 1 nt segments?
cmaffeo2's avatar
cmaffeo2 committed
697
698
699
        if len(existing_beads) == 0 or existing_beads[0].get_nt_position(self) > 0.5:
            # if len(existing_beads) > 0:            
            #     assert(existing_beads[0].get_nt_position(self) >= 0.5)
700
            b = self._generate_one_bead( self.nt_pos_to_contour(0), 0)
701
            existing_beads = [b] + existing_beads
cmaffeo2's avatar
cmaffeo2 committed
702
703

        if existing_beads[-1].get_nt_position(self)-(self.num_nts-1) < -0.5:
704
            b = self._generate_one_bead( self.nt_pos_to_contour(self.num_nts-1), 0)
705
706
707
708
709
            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
710
        last = None
cmaffeo2's avatar
cmaffeo2 committed
711

712
713
        for I in range(len(existing_beads)-1):
            eb1,eb2 = [existing_beads[i] for i in (I,I+1)]
714
            assert( eb1 is not eb2 )
715

cmaffeo2's avatar
cmaffeo2 committed
716
717
718
719
            # if np.isclose(eb1.position[2], eb2.position[2]):
            #     import pdb
            #     pdb.set_trace()

720
            # print(" %s working on %d to %d" % (self.name, eb1.position[2], eb2.position[2]))
721
722
723
724
725
726
727
728
729
730
731
732
733
            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 )
            ds = e_ds / (num_beads+1)
            nts = ds*self.num_nts
            eb1.num_nts += 0.5*nts
            eb2.num_nts += 0.5*nts

            ## 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
734
                last.make_intrahelical_neighbor(eb1)
735
736
737
            last = eb1
            for j in range(num_beads):
                s = ds*(j+1) + s0
738
                # if self.name in ("51-2","51-3"):
cmaffeo2's avatar
cmaffeo2 committed
739
                # if self.name in ("31-2",):
740
                #     print(" adding bead at {}".format(s))
741
742
                b = self._generate_one_bead(s,nts)

cmaffeo2's avatar
cmaffeo2 committed
743
                last.make_intrahelical_neighbor(b)
744
745
746
                last = b
                tmp_children.append(b)

cmaffeo2's avatar
cmaffeo2 committed
747
        last.make_intrahelical_neighbor(eb2)
748
749
750

        if eb2.parent == self:
            tmp_children.append(eb2)
cmaffeo2's avatar
cmaffeo2 committed
751
        # if self.name in ("31-2",):
752
        #     pdb.set_trace()
753
        self._rebuild_children(tmp_children)
754
755
756
757
758
759
760
761
762
763

    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 """

764
    def __init__(self, name, num_bp, start_position = np.array((0,0,0)),
765
766
                 end_position = None, 
                 segment_model = None,
cmaffeo2's avatar
cmaffeo2 committed
767
768
                 local_twist = False,
                 num_turns = None,
cmaffeo2's avatar
cmaffeo2 committed
769
770
                 start_orientation = None,
                 twist_persistence_length = 90 ):
cmaffeo2's avatar
cmaffeo2 committed
771
772
773
        
        self.helical_rise = 10.44
        self.distance_per_nt = 3.4
774
        Segment.__init__(self, name, num_bp,
775
776
777
                         start_position,
                         end_position, 
                         segment_model)
778
        self.num_bp = self.num_nts
779

cmaffeo2's avatar
cmaffeo2 committed
780
781
        self.local_twist = local_twist
        if num_turns is None:
782
783
            num_turns = float(num_bp) / self.helical_rise
        self.twist_per_nt = float(360 * num_turns) / num_bp
cmaffeo2's avatar
cmaffeo2 committed
784
785

        if start_orientation is None:
786
            start_orientation = np.eye(3) # np.array(((1,0,0),(0,1,0),(0,0,1)))
cmaffeo2's avatar
cmaffeo2 committed
787
        self.start_orientation = start_orientation
cmaffeo2's avatar
cmaffeo2 committed
788
        self.twist_persistence_length = twist_persistence_length
cmaffeo2's avatar
cmaffeo2 committed
789

790
791
        self.nicks = []

792
        self.start = self.start5 = Location( self, address=0, type_= "end5" )
793
        self.start3 = Location( self, address=0, type_ = "end3", on_fwd_strand=False )
794

795
796
        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
797
798
        # for l in (self.start5,self.start3,self.end3,self.end5):
        #     self.locations.append(l)
799

800
801
802
803
804
805
806
        ## 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

807
    ## Convenience methods
808
    ## TODO: add errors if unrealistic connections are made
809
    ## TODO: make connections automatically between unconnected strands
810
    def connect_start5(self, end3, type_="intrahelical", force_connection=False):
811
812
        if isinstance(end3, SingleStrandedSegment):
            end3 = end3.end3
813
814
        self._connect_ends( self.start5, end3, type_, force_connection = force_connection )
    def connect_start3(self, end5, type_="intrahelical", force_connection=False):
815
        if isinstance(end5, SingleStrandedSegment):
816
            end5 = end5.start5
817
818
        self._connect_ends( self.start3, end5, type_, force_connection = force_connection )
    def connect_end3(self, end5, type_="intrahelical", force_connection=False):
819
        if isinstance(end5, SingleStrandedSegment):
820
            end5 = end5.start5
821
822
        self._connect_ends( self.end3, end5, type_, force_connection = force_connection )
    def connect_end5(self, end3, type_="intrahelical", force_connection=False):
823
824
        if isinstance(end3, SingleStrandedSegment):
            end3 = end3.end3
825
        self._connect_ends( self.end5, end3, type_, force_connection = force_connection )
826

827
    def add_crossover(self, nt, other, other_nt, strands_fwd=(True,False), nt_on_5prime=True, type_="crossover"):
cmaffeo2's avatar
cmaffeo2 committed
828
829
830
831
        """ Add a crossover between two helices """
        ## Validate other, nt, other_nt
        ##   TODO

832
        if isinstance(other,SingleStrandedSegment):
cmaffeo2's avatar
cmaffeo2 committed
833
            other.add_crossover(other_nt, self, nt, strands_fwd[::-1], not nt_on_5prime)
834
        else:
835

836
837
838
            ## Create locations, connections and add to segments
            c = self.nt_pos_to_contour(nt)
            assert(c >= 0 and c <= 1)
839

840
841
842
            loc = self.get_location_at(c, strands_fwd[0])

            c = other.nt_pos_to_contour(other_nt)
cmaffeo2's avatar
cmaffeo2 committed
843
            # TODOTODO: may need to subtract or add a little depending on 3prime/5prime
844
845
            assert(c >= 0 and c <= 1)
            other_loc = other.get_location_at(c, strands_fwd[1])
846
            self._connect(other, Connection( loc, other_loc, type_=type_ ))
cmaffeo2's avatar
cmaffeo2 committed
847
848
849
850
851
852
            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
853

854
    ## Real work
855
    def _connect_ends(self, end1, end2, type_, force_connection):
856
        ## TODO remove self?
857
858
859
860
861
        ## validate the input
        for end in (end1, end2):
            assert( isinstance(end, Location) )
            assert( end.type_ in ("end3","end5") )
        assert( end1.type_ != end2.type_ )
862
        ## Create and add connection
cmaffeo2's avatar
cmaffeo2 committed
863
        if end2.type_ == "end5":
864
865
866
            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 )
867
    def _get_num_beads(self, contour, max_basepairs_per_bead, max_nucleotides_per_bead):
868
869
        # return int(contour*self.num_nts // max_basepairs_per_bead)
        return int(contour*(self.num_nts**2/(self.num_nts+1)) // max_basepairs_per_bead)
cmaffeo2's avatar
cmaffeo2 committed
870

871
872
    def _generate_one_bead(self, contour_position, nts):
        pos = self.contour_to_position(contour_position)
cmaffeo2's avatar
cmaffeo2 committed
873
        if self.local_twist:
874
            orientation = self.contour_to_orientation(contour_position)
cmaffeo2's avatar
cmaffeo2 committed
875
876
877
            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
878
            opos = pos + orientation.dot( np.array((Segment.orientation_bond.r0,0,0)) )
cmaffeo2's avatar
cmaffeo2 committed
879
880
881
882
883
            # 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,
884
                                 num_nts=nts, parent=self )
885
            bead = SegmentParticle( Segment.dsDNA_particle, pos, name="DNA",
886
887
888
                                    num_nts=nts, parent=self, 
                                    orientation_bead=o,
                                    contour_position=contour_position )
cmaffeo2's avatar
cmaffeo2 committed
889
890

        else:
891
            bead = SegmentParticle( Segment.dsDNA_particle, pos, name="DNA",
892
893
894
                                    num_nts=nts, parent=self,
                                    contour_position=contour_position )
        self._add_bead(bead)
cmaffeo2's avatar
cmaffeo2 committed
895
        return bead
896
897
898
899
900
901

class SingleStrandedSegment(Segment):

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

cmaffeo2's avatar
cmaffeo2 committed
902
    def __init__(self, name, num_nts, start_position = None,
903
904
905
                 end_position = None, 
                 segment_model = None):

cmaffeo2's avatar
cmaffeo2 committed
906
        if start_position is None: start_position = np.array((0,0,0))
907
908
909
910
911
912
        self.distance_per_nt = 5
        Segment.__init__(self, name, num_nts, 
                         start_position,
                         end_position, 
                         segment_model)

913
        self.start = self.start5 = Location( self, address=0, type_= "end5" ) # TODO change type_?
914
        self.end = self.end3 = Location( self, address=1, type_ = "end3" )
cmaffeo2's avatar
cmaffeo2 committed
915
916
        # for l in (self.start5,self.end3):
        #     self.locations.append(l)
917

918
    def connect_end3(self, end5, force_connection=False):
cmaffeo2's avatar
cmaffeo2 committed
919
        self._connect_end( end5,  _5_to_3 = True, force_connection = force_connection )
920

921
    def connect_5end(self, end3, force_connection=False): # TODO: change name or possibly deprecate
cmaffeo2's avatar
cmaffeo2 committed
922
        self._connect_end( end3,  _5_to_3 = False, force_connection = force_connection )
923
924
925
926

    def _connect_end(self, other, _5_to_3, force_connection):
        assert( isinstance(other, Location) )
        if _5_to_3 == True:
cmaffeo2's avatar
cmaffeo2 committed
927
928
929
930
            my_end = self.end3
            # assert( other.type_ == "end5" )
            if (other.type_ is not "end5"):
                print("Warning: code does not prevent connecting 3prime to 3prime, etc")
931
932
            conn = Connection( my_end, other, type_="intrahelical" )
            self._connect( other.container, conn, in_3prime_direction=True )
933
        else:
cmaffeo2's avatar
cmaffeo2 committed
934
935
936
937
            my_end = self.end5
            # assert( other.type_ == "end3" )
            if (other.type_ is not "end3"):
                print("Warning: code does not prevent connecting 3prime to 3prime, etc")
938
939
            conn = Connection( other, my_end, type_="intrahelical" )
            other.container._connect( self, conn, in_3prime_direction=True )
940

cmaffeo2's avatar
cmaffeo2 committed
941
    def add_crossover(self, nt, other, other_nt, strands_fwd=(True,False), nt_on_5prime=True):
942
943
944
945
946
947
        """ Add a crossover between two helices """
        ## Validate other, nt, other_nt
        ##   TODO
       
        ## TODO: fix direction

cmaffeo2's avatar
cmaffeo2 committed
948
949
950
951
952
953
954
955
956
957
958
        # c1 = self.nt_pos_to_contour(nt)
        # # TODOTODO
        # ## Ensure connections occur at ends, otherwise the structure doesn't make sense
        # # assert(np.isclose(c1,0) or np.isclose(c1,1))
        # assert(np.isclose(nt,0) or np.isclose(nt,self.num_nts-1))
        if nt == 0:
            c1 = 0
        elif nt == self.num_nts-1:
            c1 = 1
        else:
            raise Exception("Crossovers can only be at the ends of an ssDNA segment")
959
960
        loc = self.get_location_at(c1, True)

cmaffeo2's avatar
cmaffeo2 committed
961
962
963
964
965
966
967
        if other_nt == 0:
            c2 = 0
        elif other_nt == other.num_nts-1:
            c2 = 1
        else:
            c2 = other.nt_pos_to_contour(other_nt)

968
969
        if isinstance(other,SingleStrandedSegment):
            ## Ensure connections occur at opposing ends
cmaffeo2's avatar
cmaffeo2 committed
970
            assert(np.isclose(other_nt,0) or np.isclose(other_nt,self.num_nts-1))
971
            other_loc = other.get_location_at( c2, True )
972
973
            # if ("22-2" in (self.name, other.name)):
            #     pdb.set_trace()
cmaffeo2's avatar
cmaffeo2 committed
974
            if nt_on_5prime:
975
976
977
978
979
980
981
                self.connect_end3( other_loc )
            else:
                other.connect_end3( self )

        else:
            assert(c2 >= 0 and c2 <= 1)
            other_loc = other.get_location_at( c2, strands_fwd[1] )
cmaffeo2's avatar
cmaffeo2 committed
982
            if nt_on_5prime:
983
984
985
                self._connect(other, Connection( loc, other_loc, type_="sscrossover" ), in_3prime_direction=True )
            else:
                other._connect(self, Connection( other_loc, loc, type_="sscrossover" ), in_3prime_direction=True )
986

987
    def _get_num_beads(self, contour, max_basepairs_per_bead, max_nucleotides_per_bead):
988
989
        return int(contour*(self.num_nts**2/(self.num_nts+1)) // max_basepairs_per_bead)
        # return int(contour*self.num_nts // max_nucleotides_per_bead)
cmaffeo2's avatar
cmaffeo2 committed
990

991
992
    def _generate_one_bead(self, contour_position, nts):
        pos = self.contour_to_position(contour_position)
993
994
        b = SegmentParticle( Segment.ssDNA_particle, pos, 
                             name="NAS",
995
996
997
998
                             num_nts=nts, parent=self,
                             contour_position=contour_position )
        self._add_bead(b)
        return b
999
1000

    
cmaffeo2's avatar
cmaffeo2 committed
1001
class StrandInSegment(Group):
1002
    """ Represents a piece of an ssDNA strand within a segment """
cmaffeo2's avatar
cmaffeo2 committed
1003
1004
    
    def __init__(self, segment, start, end, is_fwd):
1005
        """ start/end should be provided expressed in nt coordinates, is_fwd tuples """
cmaffeo2's avatar
cmaffeo2 committed
1006
1007
        Group.__init__(self)
        self.num_nts = 0
1008
        # self.sequence = []
cmaffeo2's avatar