segmentmodel.py 109 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
 - develop unit test suite
38
"""
cmaffeo2's avatar
cmaffeo2 committed
39
40
class CircularDnaError(Exception):
    pass
41

cmaffeo2's avatar
cmaffeo2 committed
42
43
44
class ParticleNotConnectedError(Exception):
    pass

45
46
class Location():
    """ Site for connection within an object """
47
    def __init__(self, container, address, type_, on_fwd_strand = True):
48
        ## TODO: remove cyclic references(?)
49
        self.container = container
cmaffeo2's avatar
cmaffeo2 committed
50
        self.address = address  # represents position along contour length in segment
cmaffeo2's avatar
cmaffeo2 committed
51
        # assert( type_ in ("end3","end5") ) # TODO remove or make conditional
52
        self.on_fwd_strand = on_fwd_strand
53
54
        self.type_ = type_
        self.particle = None
55
        self.connection = None
56
        self.is_3prime_side_of_connection = None
57

58
59
        self.prev_in_strand = None
        self.next_in_strand = None
60
61
        
        self.combine = None     # some locations might be combined in bead model 
62
63
64
65
66
67
68

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

69
    def set_connection(self, connection, is_3prime_side_of_connection):
70
        self.connection = connection # TODO weakref? 
71
        self.is_3prime_side_of_connection = is_3prime_side_of_connection
72

73
74
75
76
77
78
79
    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:
80
                pos = self.container.num_nt-1
81
82
83
84
            else:
                raise
        return pos

85
86
87
88
89
90
91
    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)
        
92
93
94
95
96
97
98
99
100
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_
        
101
102
103
104
105
106
107
    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
108

109
110
111
112
113
114
    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
115
116
117
118
    def __repr__(self):
        return "<Connection {}--{}--{}]>".format( self.A, self.type_, self.B )
        

119
        
120
121
122
# class ConnectableElement(Transformable):
class ConnectableElement():
    """ Abstract base class """
cmaffeo2's avatar
cmaffeo2 committed
123
124
125
126
    def __init__(self, connection_locations=None, connections=None):
        if connection_locations is None: connection_locations = []
        if connections is None: connections = []

127
128
        ## TODO decide on names
        self.locations = self.connection_locations = connection_locations
129
130
        self.connections = connections

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

    def get_location_at(self, address, on_fwd_strand=True, new_type="crossover"):
        loc = None
144
        if (self.num_nt == 1):
145
146
147
148
149
150
151
            # 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
152
            # assert( loc is not None )
153
154
155
156
157
158
159
160
        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
161

cmaffeo2's avatar
cmaffeo2 committed
162
    def get_connections_and_locations(self, connection_type=None, exclude=()):
163
164
        """ Returns a list with each entry of the form:
            connection, location_in_self, location_in_other """
165
        type_ = connection_type
166
167
        ret = []
        for c in self.connections:
168
            if (type_ is None or c.type_ == type_) and c.type_ not in exclude:
169
                if   c.A.container is self:
170
                    ret.append( [c, c.A, c.B] )
171
                elif c.B.container is self:
172
173
                    ret.append( [c, c.B, c.A] )
                else:
174
175
                    import pdb
                    pdb.set_trace()
176
177
178
                    raise Exception("Object contains connection that fails to refer to object")
        return ret

179
    def _connect(self, other, connection, in_3prime_direction=None):
180
181
        ## TODO fix circular references        
        A,B = [connection.A, connection.B]
182
183
184
185
        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
            
186
        A.connection = B.connection = connection
187
188
        self.connections.append(connection)
        other.connections.append(connection)
189
190
191
192
193
194
        l = A.container.locations
        if A not in l: l.append(A)
        l = B.container.locations
        if B not in l: l.append(B)
        

195
196
    # def _find_connections(self, loc):
    #     return [c for c in self.connections if c.A == loc or c.B == loc]
197
198
199

class SegmentParticle(PointParticle):
    def __init__(self, type_, position, name="A", segname="A", **kwargs):
200
        self.name = name
201
202
203
204
        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
205
        self.locations = []
206
207
208

    def get_intrahelical_above(self):
        """ Returns bead directly above self """
209
        # assert( len(self.intrahelical_neighbors) <= 2 )
210
211
212
213
214
215
        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 """
216
        # assert( len(self.intrahelical_neighbors) <= 2 )
217
218
219
220
        for b in self.intrahelical_neighbors:
            if b.get_contour_position(self.parent) < self.contour_position:
                return b

cmaffeo2's avatar
cmaffeo2 committed
221
    def _neighbor_should_be_added(self,b):
222
223
224
        if type(self.parent) != type(b.parent):
            return True

cmaffeo2's avatar
cmaffeo2 committed
225
226
227
228
        c1 = self.contour_position
        c2 = b.get_contour_position(self.parent)
        if c2 < c1:
            b0 = self.get_intrahelical_below()
229
        else:
cmaffeo2's avatar
cmaffeo2 committed
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
            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:
247
248
            # assert(len(b.intrahelical_neighbors) <= 1)
            # assert(len(self.intrahelical_neighbors) <= 1)
cmaffeo2's avatar
cmaffeo2 committed
249
250
            self.intrahelical_neighbors.append(b)
            b.intrahelical_neighbors.append(self)
251

cmaffeo2's avatar
cmaffeo2 committed
252
253
254
255
256
257
258
259
260
261
    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)

262
263
264
265
    def get_contour_position(self,seg):
        if seg == self.parent:
            return self.contour_position
        else:
cmaffeo2's avatar
cmaffeo2 committed
266
267
268
269
270
            cutoff = 30*3
            target_seg = seg

            ## depth-first search
            ## TODO cache distances to nearby locations?
cmaffeo2's avatar
cmaffeo2 committed
271
            def descend_search_tree(seg, contour_in_seg, distance=0, visited_segs=None):
cmaffeo2's avatar
cmaffeo2 committed
272
                nonlocal cutoff
cmaffeo2's avatar
cmaffeo2 committed
273
                if visited_segs is None: visited_segs = []
cmaffeo2's avatar
cmaffeo2 committed
274
275
276
277
278
279
280
281

                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
282
                    return [[distance, contour_in_seg+sign*seg.nt_pos_to_contour(distance)]], [(seg, contour_in_seg, distance)]
cmaffeo2's avatar
cmaffeo2 committed
283
                if distance > cutoff:
284
                    return None,None
cmaffeo2's avatar
cmaffeo2 committed
285
286
                    
                ret_list = []
287
                hist_list = []
cmaffeo2's avatar
cmaffeo2 committed
288
                ## Find intrahelical locations in seg that we might pass through
289
290
291
292
293
                conn_locs = seg.get_connections_and_locations("intrahelical")
                if isinstance(target_seg, SingleStrandedSegment):
                    tmp = seg.get_connections_and_locations("sscrossover")
                    conn_locs = conn_locs + list(filter(lambda x: x[2].container == target_seg, tmp))
                for c,A,B in conn_locs:
cmaffeo2's avatar
cmaffeo2 committed
294
                    if B.container in visited_segs: continue
295
296
297
                    dx = seg.contour_to_nt_pos( A.address, round_nt=False ) - seg.contour_to_nt_pos( contour_in_seg, round_nt=False)
                    dx = np.abs(dx)
                    results,history = descend_search_tree( B.container, B.address,
cmaffeo2's avatar
cmaffeo2 committed
298
299
300
                                                   distance+dx, visited_segs + [seg] )
                    if results is not None:
                        ret_list.extend( results )
301
302
                        hist_list.extend( history )
                return ret_list,hist_list
cmaffeo2's avatar
cmaffeo2 committed
303

304
            results,history = descend_search_tree(self.parent, self.contour_position)
cmaffeo2's avatar
cmaffeo2 committed
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
            if results is None or len(results) == 0:
                raise Exception("Could not find location in segment") # TODO better error
            return sorted(results,key=lambda x:x[0])[0][1]

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

    def update_position(self, contour_position):
        self.contour_position = contour_position
        self.position = self.parent.contour_to_position(contour_position)
        if 'orientation_bead' in self.__dict__:
            o = self.orientation_bead
            o.contour_position = contour_position
            orientation = self.parent.contour_to_orientation(contour_position)
            if orientation is None:
                print("WARNING: local_twist is True, but orientation is None; using identity")
                orientation = np.eye(3)
            o.position = self.position + orientation.dot( np.array((Segment.orientation_bond.r0,0,0)) )
            
cmaffeo2's avatar
cmaffeo2 committed
324
325
326
    def __repr__(self):
        return "<SegmentParticle {} on {}[{:.2f}]>".format( self.name, self.parent, self.contour_position)

327
328

## TODO break this class into smaller, better encapsulated pieces
329
330
331
332
333
334
335
336
337
338
339
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
340
341
342
343
344
    orientation_particle = ParticleType("O",
                                        diffusivity = 100,
                                        mass = 300,
                                        radius = 1,
                                    )
345

cmaffeo2's avatar
cmaffeo2 committed
346
    # orientation_bond = HarmonicBond(10,2)
347
    orientation_bond = HarmonicBond(30,1.5, rRange = (0,500) )
348
349
350
351
352
353
354

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

355
    def __init__(self, name, num_nt, 
cmaffeo2's avatar
cmaffeo2 committed
356
                 start_position = None,
357
358
359
                 end_position = None, 
                 segment_model = None):

cmaffeo2's avatar
cmaffeo2 committed
360
361
        if start_position is None: start_position = np.array((0,0,0))

362
        Group.__init__(self, name, children=[])
363
        ConnectableElement.__init__(self, connection_locations=[], connections=[])
364

365
        self.resname = name
cmaffeo2's avatar
cmaffeo2 committed
366
367
368
369
370
        self.start_orientation = None
        self.twist_per_nt = 0

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

371
372
373
        self._bead_model_generation = 0    # TODO: remove?
        self.segment_model = segment_model # TODO: remove?

374
375
376
377
        self.strand_pieces = dict()
        for d in ('fwd','rev'):
            self.strand_pieces[d] = []

378
        self.num_nt = int(num_nt)
379
        if end_position is None:
380
            end_position = np.array((0,0,self.distance_per_nt*num_nt)) + start_position
381
382
383
        self.start_position = start_position
        self.end_position = end_position

384
        ## Set up interpolation for positions
385
386
387
388
        self._set_splines_from_ends()

        self.sequence = None

389
390
391
    def __repr__(self):
        return "<{} {}[{:d}]>".format( type(self), self.name, self.num_nt )

cmaffeo2's avatar
cmaffeo2 committed
392
    def set_splines(self, contours, coords):
393
394
        tck, u = interpolate.splprep( coords.T, u=contours, s=0, k=1)
        self.position_spline_params = tck
395

396
397
398
399
    def set_orientation_splines(self, contours, quaternions):
        tck, u = interpolate.splprep( quaternions.T, u=contours, s=0, k=1)
        self.quaternion_spline_params = tck

400
    def _set_splines_from_ends(self):
401
        self.quaternion_spline_params = None
cmaffeo2's avatar
cmaffeo2 committed
402
403
        coords = np.array([self.start_position, self.end_position])
        self.set_splines([0,1], coords)
404

405
406
407
408
409
    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
410

411
    def contour_to_nt_pos(self, contour_pos, round_nt=False):
412
        nt = contour_pos*(self.num_nt) - 0.5
413
        if round_nt:
cmaffeo2's avatar
cmaffeo2 committed
414
            assert( np.isclose(np.around(nt),nt) )
415
416
417
            nt = np.around(nt)
        return nt

418
    def nt_pos_to_contour(self,nt_pos):
419
        return (nt_pos+0.5)/(self.num_nt)
420

421
422
423
424
425
426
427
    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 )
428
429
        t = (t / np.linalg.norm(t,axis=0))
        return t.T
430
431
432
        

    def contour_to_orientation(self,s):
433
        assert( isinstance(s,float) or isinstance(s,int) or len(s) == 1 )   # TODO make vectorized version
434

435
436
437
438
439
        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)
440
            zAxis = np.array((0,0,1))
441
442

            if rotAxisL > 0.001:
443
444
445
                theta = np.arcsin(rotAxisL) * 180/np.pi
                if axis.dot(zAxis) < 0: theta = 180-theta
                orientation0 = rotationAboutAxis( rotAxis/rotAxisL, theta, normalizeAxis=False ).T
446
            else:
447
448
                orientation0 = np.eye(3) if axis.dot(zAxis) > 0 else \
                               rotationAboutAxis( np.array((1,0,0)), 180, normalizeAxis=False )
449
450
451
452
453
454
            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)
455

456
        return orientation
457

cmaffeo2's avatar
cmaffeo2 committed
458
    def get_contour_sorted_connections_and_locations(self,type_):
cmaffeo2's avatar
cmaffeo2 committed
459
        sort_fn = lambda c: c[1].address
cmaffeo2's avatar
cmaffeo2 committed
460
        cl = self.get_connections_and_locations(type_)
cmaffeo2's avatar
cmaffeo2 committed
461
        return sorted(cl, key=sort_fn)
462
463
464
    
    def randomize_unset_sequence(self):
        bases = list(seqComplement.keys())
465
        # bases = ['T']        ## FOR DEBUG
466
        if self.sequence is None:
467
            self.sequence = [random.choice(bases) for i in range(self.num_nt)]
468
        else:
469
            assert(len(self.sequence) == self.num_nt) # TODO move
470
471
472
            for i in range(len(self.sequence)):
                if self.sequence[i] is None:
                    self.sequence[i] = random.choice(bases)
473

cmaffeo2's avatar
cmaffeo2 committed
474
475
476
    def _get_num_beads(self, max_basepairs_per_bead, max_nucleotides_per_bead ):
        raise NotImplementedError

477
    def _generate_one_bead(self, contour_position, nts):
478
479
        raise NotImplementedError

480
    def _generate_atomic_nucleotide(self, contour_position, is_fwd, seq, scale):
cmaffeo2's avatar
cmaffeo2 committed
481
482
483
484
485
        """ 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)
486
        orientation = self.contour_to_orientation(contour_position)
cmaffeo2's avatar
cmaffeo2 committed
487

488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
        """ 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
506
        key = seq
507
508
        nt_dict = canonicalNtFwd if is_fwd else canonicalNtRev

509
        atoms = nt_dict[ key ].generate() # TODO: clone?
cmaffeo2's avatar
cmaffeo2 committed
510
                        
cmaffeo2's avatar
cmaffeo2 committed
511
        atoms.orientation = orientation.dot(atoms.orientation)
512
513
514
515
516
        if isinstance(self, SingleStrandedSegment):
            if scale is not None and scale != 1:
                for a in atoms:
                    a.position = scale*a.position
                    a.beta = 0
517
            atoms.position = pos - atoms.atoms_by_name["C1'"].collapsedPosition()
518
519
520
521
522
523
524
525
526
527
528
        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
529
530

        return atoms
531

532
533
    def add_location(self, nt, type_, on_fwd_strand=True):
        ## Create location if needed, add to segment
534
        c = self.nt_pos_to_contour(nt)
535
536
537
538
539
540
        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?
541
542
543
544
545

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

546
    def add_5prime(self, nt, on_fwd_strand=True):
547
548
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
549
        self.add_location(nt,"5prime",on_fwd_strand)
550
551

    def add_3prime(self, nt, on_fwd_strand=True):
552
553
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
554
        self.add_location(nt,"3prime",on_fwd_strand)
555

556
    def get_3prime_locations(self):
cmaffeo2's avatar
cmaffeo2 committed
557
        return sorted(self.get_locations("3prime"),key=lambda x: x.address)
558
    
cmaffeo2's avatar
cmaffeo2 committed
559
    def get_5prime_locations(self):
560
        ## TODO? ensure that data is consistent before _build_model calls
cmaffeo2's avatar
cmaffeo2 committed
561
        return sorted(self.get_locations("5prime"),key=lambda x: x.address)
cmaffeo2's avatar
cmaffeo2 committed
562

563
    def iterate_connections_and_locations(self, reverse=False):
cmaffeo2's avatar
cmaffeo2 committed
564
565
        ## connections to other segments
        cl = self.get_contour_sorted_connections_and_locations()
566
        if reverse:
cmaffeo2's avatar
cmaffeo2 committed
567
            cl = cl[::-1]
568
569
570
            
        for c in cl:
            yield c
cmaffeo2's avatar
cmaffeo2 committed
571

572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
    ## 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)

590
    ## TODO rename
591
    def get_strand_segment(self, nt_pos, is_fwd, move_at_least=0.5):
592
        """ Walks through locations, checking for crossovers """
593
594
595
596
        # if self.name in ("6-1","1-1"):
        #     import pdb
        #     pdb.set_trace()
        move_at_least = 0
597
598

        ## Iterate through locations
cmaffeo2's avatar
cmaffeo2 committed
599
        # locations = sorted(self.locations, key=lambda l:(l.address,not l.on_fwd_strand), reverse=(not is_fwd))
600
601
602
603
604
605
        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))
606
607
        # print(locations)

608
        for l in locations:
cmaffeo2's avatar
cmaffeo2 committed
609
610
611
612
            # TODOTODO probably okay
            if l.address == 0:
                pos = 0.0
            elif l.address == 1:
613
                pos = self.num_nt-1
cmaffeo2's avatar
cmaffeo2 committed
614
615
            else:
                pos = self.contour_to_nt_pos(l.address, round_nt=True)
616
617
618

            ## DEBUG

cmaffeo2's avatar
cmaffeo2 committed
619

620
            ## Skip locations encountered before our strand
621
622
623
624
625
626
627
628
            # 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
629
630
631

            ## Stop if we found the 3prime end
            if l.on_fwd_strand == is_fwd and l.type_ == "3prime":
632
                # print("  found end at",l)
633
                return pos, None, None, None, None
634
635
636
637
638
639
640
641

            ## 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:
642
643
                # print("  passing through",l)
                # print("from {}, connection {} to {}".format(nt_pos,l,B))
644
                Bpos = B.get_nt_pos()
645
                return pos, B.container, Bpos, B.on_fwd_strand, 0.5
646
647
648
                
            ## Stop at other strand crossovers so basepairs line up
            elif c.type_ == "crossover":
649
                if nt_pos == pos: continue
650
                # print("  pausing at",l)
651
                return pos, l.container, pos+(2*is_fwd-1), is_fwd, 0
652
653
654
655
656
657

        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

658
659
660
    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
661
        # TODO: include beads in connections?
662
663
664
        i = np.argmin((cs - contour_position)**2)

        return self.beads[i]
665

666
667
668
669
670
671
672
673
674
    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 ))

675
676
    def get_all_consecutive_beads(self, number):
        assert(number >= 1)
cmaffeo2's avatar
cmaffeo2 committed
677
        ## Assume that consecutive beads in self.beads are bonded
678
        ret = []
cmaffeo2's avatar
cmaffeo2 committed
679
680
        for i in range(len(self.beads)-number+1):
            tmp = [self.beads[i+j] for j in range(0,number)]
681
            ret.append( tmp )
682
        return ret   
683

684
685
686
    def _add_bead(self,b,set_contour=False):
        if set_contour:
            b.contour_position = b.get_contour_position(self)
687
        
688
689
690
        # assert(b.parent is None)
        if b.parent is not None:
            b.parent.children.remove(b)
691
        self.add(b)
692
693
694
695
696
697
        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)
698
            self.add(o)
699
700
701
702
703
704
705
706
            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 = []
707
708

        if True:
709
710
            ## TODO: remove this if duplicates are never found 
            # print("Searching for duplicate particles...")
711
            ## Remove duplicates, preserving order
712
713
714
715
            tmp = []
            for c in new_children:
                if c not in tmp:
                    tmp.append(c)
716
                else:
717
                    print("  DUPLICATE PARTICLE FOUND!")
718
719
            new_children = tmp

720
721
722
723
724
        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)
725
726
727
728
729
            
        # 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)
730
731
        assert(len(old_children) == len(self.children))
        assert(len(old_beads) == len(self.beads))
732

733

cmaffeo2's avatar
cmaffeo2 committed
734
    def _generate_beads(self, bead_model, max_basepairs_per_bead, max_nucleotides_per_bead):
735

736
        """ Generate beads (positions, types, etc) and bonds, angles, dihedrals, exclusions """
cmaffeo2's avatar
cmaffeo2 committed
737
        ## TODO: decide whether to remove bead_model argument
738
        ##       (currently unused)
cmaffeo2's avatar
cmaffeo2 committed
739

740
        ## First find points between-which beads must be generated
741
742
743
        # 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
744
745
        # if self.name == "S001":
        #     pdb.set_trace()
746

cmaffeo2's avatar
cmaffeo2 committed
747
748
        existing_beads0 = {l.particle for l in self.locations if l.particle is not None}
        existing_beads = sorted( list(existing_beads0), key=lambda b: b.get_contour_position(self) )
749

750
751
752
        # if self.num_nt == 1 and all([l.particle is not None for l in self.locations]):
        #     pdb.set_trace()
        #     return
753

754
755
756
757
        for b in existing_beads:
            assert(b.parent is not None)

        ## Add ends if they don't exist yet
758
        ## TODOTODO: test 1 nt segments?
759
        if len(existing_beads) == 0 or existing_beads[0].get_nt_position(self) >= 0.5:
cmaffeo2's avatar
cmaffeo2 committed
760
761
            # if len(existing_beads) > 0:            
            #     assert(existing_beads[0].get_nt_position(self) >= 0.5)
762
            b = self._generate_one_bead( self.nt_pos_to_contour(0), 0)
763
            existing_beads = [b] + existing_beads
cmaffeo2's avatar
cmaffeo2 committed
764

cmaffeo2's avatar
cmaffeo2 committed
765
        if existing_beads[-1].get_nt_position(self)-(self.num_nt-1) < -0.5 or len(existing_beads)==1:
766
            b = self._generate_one_bead( self.nt_pos_to_contour(self.num_nt-1), 0)
767
768
769
770
771
            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
772
        last = None
cmaffeo2's avatar
cmaffeo2 committed
773

774
775
        for I in range(len(existing_beads)-1):
            eb1,eb2 = [existing_beads[i] for i in (I,I+1)]
776
            assert( eb1 is not eb2 )
777

cmaffeo2's avatar
cmaffeo2 committed
778
779
780
781
            # if np.isclose(eb1.position[2], eb2.position[2]):
            #     import pdb
            #     pdb.set_trace()

782
            # print(" %s working on %d to %d" % (self.name, eb1.position[2], eb2.position[2]))
783
784
            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
785
786
787
788
789
790

            ## 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

791
            ds = e_ds / (num_beads+1)
792
793
794
            nts = ds*self.num_nt
            eb1.num_nt += 0.5*nts
            eb2.num_nt += 0.5*nts
795
796
797
798
799
800
801

            ## 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
802
                last.make_intrahelical_neighbor(eb1)
803
804
805
            last = eb1
            for j in range(num_beads):
                s = ds*(j+1) + s0
806
                # if self.name in ("51-2","51-3"):
cmaffeo2's avatar
cmaffeo2 committed
807
                # if self.name in ("31-2",):
808
                #     print(" adding bead at {}".format(s))
809
810
                b = self._generate_one_bead(s,nts)

cmaffeo2's avatar
cmaffeo2 committed
811
                last.make_intrahelical_neighbor(b)
812
813
814
                last = b
                tmp_children.append(b)

cmaffeo2's avatar
cmaffeo2 committed
815
        last.make_intrahelical_neighbor(eb2)
816
817
818

        if eb2.parent == self:
            tmp_children.append(eb2)
cmaffeo2's avatar
cmaffeo2 committed
819
        # if self.name in ("31-2",):
820
        #     pdb.set_trace()
821
        self._rebuild_children(tmp_children)
822
823
824
825
826
827
828
829
830
831

    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 """

832
    def __init__(self, name, num_bp, start_position = np.array((0,0,0)),
833
834
                 end_position = None, 
                 segment_model = None,
cmaffeo2's avatar
cmaffeo2 committed
835
836
                 local_twist = False,
                 num_turns = None,
cmaffeo2's avatar
cmaffeo2 committed
837
838
                 start_orientation = None,
                 twist_persistence_length = 90 ):
cmaffeo2's avatar
cmaffeo2 committed
839
840
841
        
        self.helical_rise = 10.44
        self.distance_per_nt = 3.4
842
        Segment.__init__(self, name, num_bp,
843
844
845
                         start_position,
                         end_position, 
                         segment_model)
846
        self.num_bp = self.num_nt
847

cmaffeo2's avatar
cmaffeo2 committed
848
849
        self.local_twist = local_twist
        if num_turns is None:
850
851
            num_turns = float(num_bp) / self.helical_rise
        self.twist_per_nt = float(360 * num_turns) / num_bp
cmaffeo2's avatar
cmaffeo2 committed
852
853

        if start_orientation is None:
854
            start_orientation = np.eye(3) # np.array(((1,0,0),(0,1,0),(0,0,1)))
cmaffeo2's avatar
cmaffeo2 committed
855
        self.start_orientation = start_orientation
cmaffeo2's avatar
cmaffeo2 committed
856
        self.twist_persistence_length = twist_persistence_length
cmaffeo2's avatar
cmaffeo2 committed
857

858
859
        self.nicks = []

860
        self.start = self.start5 = Location( self, address=0, type_= "end5" )
861
        self.start3 = Location( self, address=0, type_ = "end3", on_fwd_strand=False )
862

863
864
        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
865
866
        # for l in (self.start5,self.start3,self.end3,self.end5):
        #     self.locations.append(l)
867

868
869
870
871
872
873
874
        ## 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

875
    ## Convenience methods
876
    ## TODO: add errors if unrealistic connections are made
877
    ## TODO: make connections automatically between unconnected strands
878
    def connect_start5(self, end3, type_="intrahelical", force_connection=False):
879
880
        if isinstance(end3, SingleStrandedSegment):
            end3 = end3.end3
881
882
        self._connect_ends( self.start5, end3, type_, force_connection = force_connection )
    def connect_start3(self, end5, type_="intrahelical", force_connection=False):
883
        if isinstance(end5, SingleStrandedSegment):
884
            end5 = end5.start5
885
886
        self._connect_ends( self.start3, end5, type_, force_connection = force_connection )
    def connect_end3(self, end5, type_="intrahelical", force_connection=False):
887
        if isinstance(end5, SingleStrandedSegment):
888
            end5 = end5.start5
889
890
        self._connect_ends( self.end3, end5, type_, force_connection = force_connection )
    def connect_end5(self, end3, type_="intrahelical", force_connection=False):
891
892
        if isinstance(end3, SingleStrandedSegment):
            end3 = end3.end3
893
        self._connect_ends( self.end5, end3, type_, force_connection = force_connection )
894

895
    def add_crossover(self, nt, other, other_nt, strands_fwd=(True,False), nt_on_5prime=True, type_="crossover"):
cmaffeo2's avatar
cmaffeo2 committed
896
897
898
899
        """ Add a crossover between two helices """
        ## Validate other, nt, other_nt
        ##   TODO

900
        if isinstance(other,SingleStrandedSegment):
cmaffeo2's avatar
cmaffeo2 committed
901
            other.add_crossover(other_nt, self, nt, strands_fwd[::-1], not nt_on_5prime)
902
        else:
903

904
905
906
            ## Create locations, connections and add to segments
            c = self.nt_pos_to_contour(nt)
            assert(c >= 0 and c <= 1)
907

908
909
910
            loc = self.get_location_at(c, strands_fwd[0])

            c = other.nt_pos_to_contour(other_nt)
cmaffeo2's avatar
cmaffeo2 committed
911
            # TODOTODO: may need to subtract or add a little depending on 3prime/5prime
912
913
            assert(c >= 0 and c <= 1)
            other_loc = other.get_location_at(c, strands_fwd[1])
914
            self._connect(other, Connection( loc, other_loc, type_=type_ ))
cmaffeo2's avatar
cmaffeo2 committed
915
916
917
918
919
920
            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
921

922
    ## Real work
923
    def _connect_ends(self, end1, end2, type_, force_connection):
924
        debug = False
925
        ## TODO remove self?
926
927
928
929
930
        ## validate the input
        for end in (end1, end2):
            assert( isinstance(end, Location) )
            assert( end.type_ in ("end3","end5") )
        assert( end1.type_ != end2.type_ )
931
932
933

        ## Remove other connections involving these points
        if end1.connection is not None:
934
            if debug: print("WARNING: reconnecting {}".format(end1))
935
936
            end1.connection.delete()
        if end2.connection is not None:
937
            if debug: print("WARNING: reconnecting {}".format(end2))
938
939
            end2.connection.delete()

940
        ## Create and add connection
cmaffeo2's avatar
cmaffeo2 committed
941
        if end2.type_ == "end5":
942
943
944
            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 )
945
    def _get_num_beads(self, contour, max_basepairs_per_bead, max_nucleotides_per_bead):
946
947
        # 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)