segmentmodel.py 78 KB
Newer Older
1
import numpy as np
2
import random
3
from arbdmodel import PointParticle, ParticleType, Group, ArbdModel
4
from coords import rotationAboutAxis, quaternion_from_matrix, quaternion_to_matrix
5
6
7
8
from nonbonded import *
from copy import copy, deepcopy
from nbPot import nbDnaScheme

cmaffeo2's avatar
cmaffeo2 committed
9
10
from scipy.special import erf
import scipy.optimize as opt
11
from scipy import interpolate
cmaffeo2's avatar
cmaffeo2 committed
12

cmaffeo2's avatar
cmaffeo2 committed
13
14
15
from CanonicalNucleotideAtoms import canonicalNtFwd, canonicalNtRev, seqComplement
from CanonicalNucleotideAtoms import enmTemplateHC, enmTemplateSQ, enmCorrectionsHC

16
# import pdb
17
"""
cmaffeo2's avatar
cmaffeo2 committed
18
TODO:
cmaffeo2's avatar
cmaffeo2 committed
19
 + fix handling of crossovers for atomic representation
cmaffeo2's avatar
cmaffeo2 committed
20
 + map to atomic representation
21
22
    + add nicks
    - transform ssDNA nucleotides 
cmaffeo2's avatar
cmaffeo2 committed
23
24
    - shrink ssDNA
    - shrink dsDNA backbone
25
    + make orientation continuous
cmaffeo2's avatar
cmaffeo2 committed
26
    - sequence
27
    - handle circular dna
28
 + ensure crossover bead potentials aren't applied twice 
29
 + remove performance bottlenecks
30
31
 - test for large systems
 - assign sequence
32
 - ENM
33
34
 - rework Location class 
 - remove recursive calls
35
 - document
36
 - add unit test of helices connected to themselves
37
38
39
40
"""

class Location():
    """ Site for connection within an object """
41
    def __init__(self, container, address, type_, on_fwd_strand = True):
42
        ## TODO: remove cyclic references(?)
43
        self.container = container
44
        self.address = address  # represents position along contour length in segments
cmaffeo2's avatar
cmaffeo2 committed
45
        # assert( type_ in ("end3","end5") ) # TODO remove or make conditional
46
        self.on_fwd_strand = on_fwd_strand
47
48
        self.type_ = type_
        self.particle = None
49
        self.connection = None
50
        self.is_3prime_side_of_connection = None
51

52
53
        self.prev_in_strand = None
        self.next_in_strand = None
54
55
        
        self.combine = None     # some locations might be combined in bead model 
56
57
58
59
60
61
62

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

63
    def set_connection(self, connection, is_3prime_side_of_connection):
64
        self.connection = connection # TODO weakref? 
65
        self.is_3prime_side_of_connection = is_3prime_side_of_connection
66
67
68
69
70
71
72
73
74
75

    def __repr__(self):
        if self.on_fwd_strand:
            on_fwd = "on_fwd_strand"
        else:
            on_fwd = "on_rev_strand"
        # return "<Location in {} at contour {} {} with connection {}>".format( self.container.name, self.address, self.on_fwd_strand, self.connection )
        # return "<Location {} in {} at contour {} {} with connection {}>".format( self.type_, self.container.name, self.address, on_fwd, self.connection )
        return "<Location {}.{}[{:.2f},{:d}]>".format( self.container.name, self.type_, self.address, self.on_fwd_strand)
        
76
77
78
79
80
81
82
83
84
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_
        
85
86
87
88
89
90
91
92
    def other(self, location):
        if location is self.A:
            return self.B
        elif location is self.B:
            return self.A
        else:
            raise Exception("OutOfBoundsError")
        
93
94
95
# class ConnectableElement(Transformable):
class ConnectableElement():
    """ Abstract base class """
96
97
98
99
    ## TODO: eliminate mutable default arguments
    def __init__(self, connection_locations=[], connections=[]):
        ## TODO decide on names
        self.locations = self.connection_locations = connection_locations
100
101
        self.connections = connections

102
103
104
105
106
107
108
109
110
    def get_locations(self, type_=None, exclude=[]):
        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] ) )
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
        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
            assert( loc is not None )
        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
132
133

    def get_connections_and_locations(self, connection_type=None, exclude=[]):
134
135
        """ Returns a list with each entry of the form:
            connection, location_in_self, location_in_other """
136
        type_ = connection_type
137
138
        ret = []
        for c in self.connections:
139
            if (type_ is None or c.type_ == type_) and c.type_ not in exclude:
140
                if   c.A.container is self:
141
                    ret.append( [c, c.A, c.B] )
142
                elif c.B.container is self:
143
144
                    ret.append( [c, c.B, c.A] )
                else:
145
146
                    import pdb
                    pdb.set_trace()
147
148
149
                    raise Exception("Object contains connection that fails to refer to object")
        return ret

150
    def _connect(self, other, connection, in_3prime_direction=None):
151
152
        ## TODO fix circular references        
        A,B = [connection.A, connection.B]
153
154
155
156
        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
            
157
        A.connection = B.connection = connection
158
159
        self.connections.append(connection)
        other.connections.append(connection)
160
161
162
163
164
165
        l = A.container.locations
        if A not in l: l.append(A)
        l = B.container.locations
        if B not in l: l.append(B)
        

166
167
    # def _find_connections(self, loc):
    #     return [c for c in self.connections if c.A == loc or c.B == loc]
168
169
170

class SegmentParticle(PointParticle):
    def __init__(self, type_, position, name="A", segname="A", **kwargs):
171
        self.name = name
172
173
174
175
176
        self.contour_position = None
        PointParticle.__init__(self, type_, position, name=name, segname=segname, **kwargs)
        self.intrahelical_neighbors = []
        self.other_neighbors = []

177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
    # def get_contour_position(self,seg):
    #     assert( isinstance(seg,Segment) )
    #     if seg == self.parent:
    #         return self.contour_position
    #     else:
    #         ## TODO replace with something more elegant
    #         for c,A,B in self.parent.get_connections_and_locations():
    #             if A.particle is self and B.container is seg:
    #                 nt = np.abs( (self.contour_position - A.address)*(A.container.num_nts-1) )
    #                 if B.address < 0.5:
    #                     return B.address-nt/(seg.num_nts-1)
    #                 else:
    #                     return B.address+nt/(seg.num_nts-1)
    #         ## ERROR
    #         print("")
    #         for c,A,B in self.parent.get_connections_and_locations():
    #             print("  ",c.type_)
    #             print(A,B)
    #             print(A.particle,self)
    #             print(B.container,seg)
    #         print("")
    #         import pdb
    #         pdb.set_trace()
    #         raise Exception("Did not find location for particle {} in Segment {}".format(self,seg))

    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
        

    def get_nt_position(self,seg):
        if seg == self.parent:
            return seg.contour_to_nt_pos(self.contour_position)
        else:
cmaffeo2's avatar
cmaffeo2 committed
221
222
            cl = [e for e in self.parent.get_connections_and_locations() if e[2].container is seg]
            dc = [(self.contour_position - A.address)**2 for c,A,B in cl]
223
224
225
226
227

            if len(dc) == 0:
                pdb.set_trace()

            i = np.argmin(dc)
cmaffeo2's avatar
cmaffeo2 committed
228
            c,A,B = cl[i]
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
            ## TODO: generalize, removing np.abs and conditional 
            delta_nt = np.abs( A.container.contour_to_nt_pos(self.contour_position - A.address) )
            B_nt_pos = seg.contour_to_nt_pos(B.address)
            if B.address < 0.5:
                return B_nt_pos-delta_nt
            else:
                return B_nt_pos+delta_nt

    def get_contour_position_old(self,seg):
        if seg == self.parent:
            return self.contour_position
        else:
            cl = [e for e in self.parent.get_connections_and_locations() in B.container is seg]
            dc = [(self.contour_position - A.address)**2 for c,A,B in e]

            if len(dc) == 0:
                pdb.set_trace()

            i = np.argmin(dc)

            nt = np.abs( (self.contour_position - A.address)*(A.container.num_nts-1) )
            if B.address < 0.5:
                return seg.nt_pos_to_contour(B.address-nt)
            else:
                return seg.nt_pos_to_contour(B.address+nt)

255
256
257
258
    def get_contour_position(self,seg):
        if seg == self.parent:
            return self.contour_position
        else:
259
260
            nt_pos = self.get_nt_position(seg)
            return seg.nt_pos_to_contour(nt_pos)
261
262

## TODO break this class into smaller, better encapsulated pieces
263
264
265
266
267
268
269
270
271
272
273
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
274
275
276
277
278
    orientation_particle = ParticleType("O",
                                        diffusivity = 100,
                                        mass = 300,
                                        radius = 1,
                                    )
279

cmaffeo2's avatar
cmaffeo2 committed
280
    # orientation_bond = HarmonicBond(10,2)
281
    orientation_bond = HarmonicBond(30,1.5, rRange = (0,500) )
282
283
284
285
286
287
288
289
290
291
292
293
294

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

    def __init__(self, name, num_nts, 
                 start_position = np.array((0,0,0)),
                 end_position = None, 
                 segment_model = None):

        Group.__init__(self, name, children=[])
295
        ConnectableElement.__init__(self, connection_locations=[], connections=[])
296

297
        self.resname = name
cmaffeo2's avatar
cmaffeo2 committed
298
299
300
301
302
        self.start_orientation = None
        self.twist_per_nt = 0

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

303
304
305
        self._bead_model_generation = 0    # TODO: remove?
        self.segment_model = segment_model # TODO: remove?

cmaffeo2's avatar
cmaffeo2 committed
306
        self.num_nts = int(num_nts)
307
308
309
310
311
        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

312
313
314
315
        ## 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
316
317
        
        self.sequence = None
318

319
320
321
322
323
    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
324

325
326
327
328
329
330
331
    def contour_to_nt_pos(self, contour_pos, round_nt=False):
        nt = contour_pos*(self.num_nts-1)
        if round_nt:
            assert( (np.around(nt) - nt)**2 < 1e-3 )
            nt = np.around(nt)
        return nt

332
    def nt_pos_to_contour(self,nt_pos):
333
334
335
336
337
        if self.num_nts == 1:
            assert(nt_pos == 0)
            return 0
        else:
            return nt_pos/(self.num_nts-1)
338

339
340
341
342
343
344
345
    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 )
346
347
        t = (t / np.linalg.norm(t,axis=0))
        return t.T
348
349
350
        

    def contour_to_orientation(self,s):
351
352
        assert( isinstance(s,float) or isinstance(s,int) or len(s) == 1 )   # TODO make vectorized version
        orientation = None
353
354
355
356
        if self.start_orientation is not None:
            # axis = self.start_orientation.dot( np.array((0,0,1)) )
            if self.quaternion_spline_params is None:
                axis = self.contour_to_tangent(s)
357
                orientation = rotationAboutAxis( axis, self.twist_per_nt*self.contour_to_nt_pos(s), normalizeAxis=True )
358
359
360
361
362
            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)
        return orientation
363

cmaffeo2's avatar
cmaffeo2 committed
364
    def get_contour_sorted_connections_and_locations(self,type_):
cmaffeo2's avatar
cmaffeo2 committed
365
        sort_fn = lambda c: c[1].address
cmaffeo2's avatar
cmaffeo2 committed
366
        cl = self.get_connections_and_locations(type_)
cmaffeo2's avatar
cmaffeo2 committed
367
        return sorted(cl, key=sort_fn)
368
369
370
    
    def randomize_unset_sequence(self):
        bases = list(seqComplement.keys())
371
        # bases = ['T']        ## FOR DEBUG
372
373
374
375
376
377
378
        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)
379

cmaffeo2's avatar
cmaffeo2 committed
380
381
382
    def _get_num_beads(self, max_basepairs_per_bead, max_nucleotides_per_bead ):
        raise NotImplementedError

383
    def _generate_one_bead(self, contour_position, nts):
384
385
        raise NotImplementedError

386
    def _generate_atomic_nucleotide(self, contour_position, is_fwd, seq, scale):
cmaffeo2's avatar
cmaffeo2 committed
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
        """ 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
403
                ## TODO: improve placement of ssDNA
cmaffeo2's avatar
cmaffeo2 committed
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
                # 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
422
        atoms = nt_dict[ key ].generate() # TODO: clone?
cmaffeo2's avatar
cmaffeo2 committed
423
                        
cmaffeo2's avatar
cmaffeo2 committed
424
        atoms.orientation = orientation.dot(atoms.orientation)
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
        if isinstance(self, SingleStrandedSegment):
            if scale is not None and scale != 1:
                for a in atoms:
                    a.position = scale*a.position
                    a.beta = 0
            atoms.position = pos - atoms.atoms_by_name["C1'"].collapsed_position()
        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
442
443

        return atoms
444

445
446
    def add_location(self, nt, type_, on_fwd_strand=True):
        ## Create location if needed, add to segment
447
        c = self.nt_pos_to_contour(nt)
448
449
450
451
452
453
454
        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):
455
        self.add_location(nt,"5prime",on_fwd_strand)
456
457

    def add_3prime(self, nt, on_fwd_strand=True):
458
        self.add_location(nt,"3prime",on_fwd_strand)
459

460
461
462
    def get_3prime_locations(self):
        return self.get_locations("3prime")
    
cmaffeo2's avatar
cmaffeo2 committed
463
    def get_5prime_locations(self):
464
465
        ## TODO? ensure that data is consistent before _build_model calls
        return self.get_locations("5prime")
cmaffeo2's avatar
cmaffeo2 committed
466

467
    def iterate_connections_and_locations(self, reverse=False):
cmaffeo2's avatar
cmaffeo2 committed
468
469
        ## connections to other segments
        cl = self.get_contour_sorted_connections_and_locations()
470
        if reverse:
cmaffeo2's avatar
cmaffeo2 committed
471
            cl = cl[::-1]
472
473
474
            
        for c in cl:
            yield c
cmaffeo2's avatar
cmaffeo2 committed
475

476
    ## TODO rename
477
    def get_strand_segment(self, nt_pos, is_fwd, move_at_least=0.5):
478
        """ Walks through locations, checking for crossovers """
479
480
481
482
        # if self.name in ("6-1","1-1"):
        #     import pdb
        #     pdb.set_trace()
        move_at_least = 0
483
484
485

        ## Iterate through locations
        locations = sorted(self.locations, key=lambda l:(l.address,not l.on_fwd_strand), reverse=(not is_fwd))
486
487
        # print(locations)

488
        for l in locations:
489
            pos = self.contour_to_nt_pos(l.address, round_nt=True)
490
491
492
493

            ## DEBUG

            ## Skip locations encountered before our strand
494
495
496
497
498
499
500
501
            # 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
502
503
504

            ## Stop if we found the 3prime end
            if l.on_fwd_strand == is_fwd and l.type_ == "3prime":
505
506
                print("  found end at",l)
                return pos, None, None, None, None
507
508
509
510
511
512
513
514

            ## 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:
515
516
517
518
                print("  passing through",l)
                print("from {}, connection {} to {}".format(nt_pos,l,B))
                Bpos = B.container.contour_to_nt_pos(B.address, round_nt=True)
                return pos, B.container, Bpos, B.on_fwd_strand, 0.5
519
520
521
                
            ## Stop at other strand crossovers so basepairs line up
            elif c.type_ == "crossover":
522
523
524
                if nt_pos == pos: continue
                print("  pausing at",l)
                return pos, l.container, pos+(2*is_fwd-1), is_fwd, 0
525

526
527
        import pdb
        pdb.set_trace()
528
529
530
531
532
533
534
        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


    def get_end_of_strand_old(self, contour_pos, is_fwd):
535
536
537
538
539
        """ Walks through locations, checking for crossovers """

        ## Iterate through locations
        # for l in self.locations:
        def loc_iter():
540
541
542
543
544
545
546
            locations = sorted(self.locations, key=lambda l:(l.address,not l.on_fwd_strand), reverse=(not is_fwd))
            # if is_fwd:
            for l in locations:
                yield l
            # else:
            #     for l in locations[::-1]:
            #         yield l
547
548
549
550
551
552
553
554
555
556
557
558
559
560
            
        for l in loc_iter():
            # if l.particle is None:
            #     pos = l.address
            # else:
            #     pos = l.particle.get_contour_position()          
            pos = l.address

            ## DEBUG
            # if self.name == "1-0" and is_fwd == False:
            #     import pdb
            #     pdb.set_trace()

            ## Skip locations encountered before our strand
cmaffeo2's avatar
cmaffeo2 committed
561
562
563
564
            if is_fwd:
                if pos <= contour_pos: continue
            elif pos >= contour_pos: continue

565
566
            # print("  ?",l)
            
567
568
            ## Stop if we found the 3prime end
            if l.on_fwd_strand == is_fwd and l.type_ == "3prime":
569
                return pos, None, None, None
570
571
572
573
574
575
576
577

            ## 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:
578
                # print("  passing through",l)
579
580
581
582
583
                # print("from {}, connection {} to {}".format(contour_pos,l,B))
                return pos, B.container, B.address, B.on_fwd_strand
                
            ## Stop at other strand crossovers so basepairs line up
            elif c.type_ == "crossover":
584
                # print("  pausing at",l)
585
586
587
                # print("pausing at {}".format(l))
                return pos, l.container, pos, is_fwd

588
589
        raise Exception("Shouldn't be here")
        # print("Shouldn't be here")
590
591
        ## Made it to the end of the segment without finding a connection
        return 1*is_fwd, None, None, None
cmaffeo2's avatar
cmaffeo2 committed
592
        
593
594
595
    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
596
        # TODO: include beads in connections?
597
598
599
        i = np.argmin((cs - contour_position)**2)

        return self.beads[i]
600
601
602

    def get_all_consecutive_beads(self, number):
        assert(number >= 1)
cmaffeo2's avatar
cmaffeo2 committed
603
        ## Assume that consecutive beads in self.beads are bonded
604
        ret = []
cmaffeo2's avatar
cmaffeo2 committed
605
606
        for i in range(len(self.beads)-number+1):
            tmp = [self.beads[i+j] for j in range(0,number)]
607
            ret.append( tmp )
608
        return ret   
609

610
611
612
    def _add_bead(self,b,set_contour=False):
        if set_contour:
            b.contour_position = b.get_contour_position(self)
613
        
614
615
616
        # assert(b.parent is None)
        if b.parent is not None:
            b.parent.children.remove(b)
617
        self.add(b)
618
619
620
621
622
623
        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)
624
            self.add(o)
625
626
627
628
629
630
631
632
            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 = []
633
634
635

        if True:
            print("WARNING: DEBUG")
636
            ## Remove duplicates, preserving order
637
638
639
640
            tmp = []
            for c in new_children:
                if c not in tmp:
                    tmp.append(c)
641
642
                else:
                    print("  duplicate particle found!")
643
644
            new_children = tmp

645
646
647
648
649
        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)
650
651
652
653
654
            
        # 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)
655
656
        assert(len(old_children) == len(self.children))
        assert(len(old_beads) == len(self.beads))
657

658

cmaffeo2's avatar
cmaffeo2 committed
659
    def _generate_beads(self, bead_model, max_basepairs_per_bead, max_nucleotides_per_bead):
660

661
        """ Generate beads (positions, types, etc) and bonds, angles, dihedrals, exclusions """
cmaffeo2's avatar
cmaffeo2 committed
662
        ## TODO: decide whether to remove bead_model argument
663
        ##       (currently unused)
cmaffeo2's avatar
cmaffeo2 committed
664

665
        ## First find points between-which beads must be generated
666
667
668
669
670
671
672
673
        # 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]
        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()
674
675
676
677
        for b in existing_beads:
            assert(b.parent is not None)

        ## Add ends if they don't exist yet
678
        ## TODOTODO: test 1 nt segments?
679
        if len(existing_beads) == 0 or existing_beads[0].get_contour_position(self) > 0:
680
            if len(existing_beads) > 0:            
681
                assert(existing_beads[0].get_nt_position(self) >= 0.5)
682

683
684
685
            b = self._generate_one_bead(0, 0)
            existing_beads = [b] + existing_beads
        if existing_beads[-1].get_contour_position(self) < 1:
686
687
            # assert((1-existing_beads[0].get_contour_position(self))*(self.num_nts-1) >= 0.5)
            assert(self.num_nts-1-existing_beads[0].get_nt_position(self) >= 0.5)
688
689
690
691
692
693
            b = self._generate_one_bead(1, 0)
            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
694
        last = None
695
696
        for I in range(len(existing_beads)-1):
            eb1,eb2 = [existing_beads[i] for i in (I,I+1)]
697
698
699
            if eb1 is eb2:
                pdb.set_trace()
            assert( eb1 is not eb2 )
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716

            # print(" %s working on %d to %d" % (self.name, eb1.position[2], eb2.position[2]))
            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:
                last.intrahelical_neighbors.append(eb1)
                eb1.intrahelical_neighbors.append(last)
717
718
                assert(len(last.intrahelical_neighbors) <= 2)
                assert(len(eb1.intrahelical_neighbors) <= 2)
719
720
721
722
723
724
725
            last = eb1
            for j in range(num_beads):
                s = ds*(j+1) + s0
                b = self._generate_one_bead(s,nts)

                last.intrahelical_neighbors.append(b)
                b.intrahelical_neighbors.append(last)
726
727
                assert(len(last.intrahelical_neighbors) <= 2)
                assert(len(b.intrahelical_neighbors) <= 2)
728
729
730
731
732
                last = b
                tmp_children.append(b)

        last.intrahelical_neighbors.append(eb2)
        eb2.intrahelical_neighbors.append(last)
733
734
        assert(len(last.intrahelical_neighbors) <= 2)
        assert(len(eb2.intrahelical_neighbors) <= 2)
735
736
737
738

        if eb2.parent == self:
            tmp_children.append(eb2)
        self._rebuild_children(tmp_children)
739
740
741
742
743
744
745
746
747
748
749
750
751

    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 """

    def __init__(self, name, num_nts, start_position = np.array((0,0,0)),
                 end_position = None, 
                 segment_model = None,
cmaffeo2's avatar
cmaffeo2 committed
752
753
                 local_twist = False,
                 num_turns = None,
cmaffeo2's avatar
cmaffeo2 committed
754
755
                 start_orientation = None,
                 twist_persistence_length = 90 ):
cmaffeo2's avatar
cmaffeo2 committed
756
757
758
        
        self.helical_rise = 10.44
        self.distance_per_nt = 3.4
759
760
761
762
763
        Segment.__init__(self, name, num_nts, 
                         start_position,
                         end_position, 
                         segment_model)

cmaffeo2's avatar
cmaffeo2 committed
764
765
766
767
768
769
        self.local_twist = local_twist
        if num_turns is None:
            num_turns = float(num_nts) / self.helical_rise
        self.twist_per_nt = float(360 * num_turns) / num_nts

        if start_orientation is None:
770
            start_orientation = np.eye(3) # np.array(((1,0,0),(0,1,0),(0,0,1)))
cmaffeo2's avatar
cmaffeo2 committed
771
        self.start_orientation = start_orientation
cmaffeo2's avatar
cmaffeo2 committed
772
        self.twist_persistence_length = twist_persistence_length
cmaffeo2's avatar
cmaffeo2 committed
773

774
775
        self.nicks = []

776
        self.start = self.start5 = Location( self, address=0, type_= "end5" )
777
        self.start3 = Location( self, address=0, type_ = "end3", on_fwd_strand=False )
778

779
780
781
782
        self.end = self.end3 = Location( self, address=1, type_ = "end3" )
        self.end5 = Location( self, address=1, type_= "end5", on_fwd_strand=False )
        for l in (self.start5,self.start3,self.end3,self.end5):
            self.locations.append(l)
783

784
785
786
787
788
789
790
791
792
        ## 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
        self.quaternion_spline_params = None


793
    ## Convenience methods
794
    ## TODO: add errors if unrealistic connections are made
795
    ## TODO: make connections automatically between unconnected strands
796
    def connect_start5(self, end3, type_="intrahelical", force_connection=False):
797
798
        if isinstance(end3, SingleStrandedSegment):
            end3 = end3.end3
799
800
        self._connect_ends( self.start5, end3, type_, force_connection = force_connection )
    def connect_start3(self, end5, type_="intrahelical", force_connection=False):
801
        if isinstance(end5, SingleStrandedSegment):
802
            end5 = end5.start5
803
804
        self._connect_ends( self.start3, end5, type_, force_connection = force_connection )
    def connect_end3(self, end5, type_="intrahelical", force_connection=False):
805
        if isinstance(end5, SingleStrandedSegment):
806
            end5 = end5.start5
807
808
        self._connect_ends( self.end3, end5, type_, force_connection = force_connection )
    def connect_end5(self, end3, type_="intrahelical", force_connection=False):
809
810
        if isinstance(end3, SingleStrandedSegment):
            end3 = end3.end3
811
        self._connect_ends( self.end5, end3, type_, force_connection = force_connection )
812

813
    def add_crossover(self, nt, other, other_nt, strands_fwd=[True,False]):
cmaffeo2's avatar
cmaffeo2 committed
814
815
816
817
        """ Add a crossover between two helices """
        ## Validate other, nt, other_nt
        ##   TODO

818
819
820
        if isinstance(other,SingleStrandedSegment):
            other.add_crossover(other_nt, self, nt, strands_fwd[::-1])
        else:
821

822
823
824
            ## Create locations, connections and add to segments
            c = self.nt_pos_to_contour(nt)
            assert(c >= 0 and c <= 1)
825

826
827
828
829
830
831
832
833
834
            loc = self.get_location_at(c, strands_fwd[0])

            c = other.nt_pos_to_contour(other_nt)
            assert(c >= 0 and c <= 1)
            other_loc = other.get_location_at(c, strands_fwd[1])
            self._connect(other, Connection( loc, other_loc, type_="crossover" ))
            loc.is_3prime_side_of_connection = not strands_fwd[0]
            other_loc.is_3prime_side_of_connection = not strands_fwd[1]
            
cmaffeo2's avatar
cmaffeo2 committed
835

836
    ## Real work
837
    def _connect_ends(self, end1, end2, type_, force_connection):
838
        ## TODO remove self?
839
840
841
842
843
        ## validate the input
        for end in (end1, end2):
            assert( isinstance(end, Location) )
            assert( end.type_ in ("end3","end5") )
        assert( end1.type_ != end2.type_ )
844
        ## Create and add connection
845
846
847
848
        if end2.type_ == "end3":
            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 )
849
850
    def _get_num_beads(self, contour, max_basepairs_per_bead, max_nucleotides_per_bead):
        return int(contour*self.num_nts // max_basepairs_per_bead)
cmaffeo2's avatar
cmaffeo2 committed
851

852
853
    def _generate_one_bead(self, contour_position, nts):
        pos = self.contour_to_position(contour_position)
cmaffeo2's avatar
cmaffeo2 committed
854
        if self.local_twist:
855
            orientation = self.contour_to_orientation(contour_position)
cmaffeo2's avatar
cmaffeo2 committed
856
857
858
            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
859
            opos = pos + orientation.dot( np.array((Segment.orientation_bond.r0,0,0)) )
860
861
            o = SegmentParticle( Segment.orientation_particle, opos, nts,
                                 num_nts=nts, parent=self )
862
            bead = SegmentParticle( Segment.dsDNA_particle, pos, name="DNA",
863
864
865
                                    num_nts=nts, parent=self, 
                                    orientation_bead=o,
                                    contour_position=contour_position )
cmaffeo2's avatar
cmaffeo2 committed
866
867

        else:
868
            bead = SegmentParticle( Segment.dsDNA_particle, pos, name="DNA",
869
870
871
                                    num_nts=nts, parent=self,
                                    contour_position=contour_position )
        self._add_bead(bead)
cmaffeo2's avatar
cmaffeo2 committed
872
        return bead
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888

class SingleStrandedSegment(Segment):

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

    def __init__(self, name, num_nts, start_position = np.array((0,0,0)),
                 end_position = None, 
                 segment_model = None):

        self.distance_per_nt = 5
        Segment.__init__(self, name, num_nts, 
                         start_position,
                         end_position, 
                         segment_model)

889
        self.start = self.start5 = Location( self, address=0, type_= "end5" ) # TODO change type_?
890
        self.end = self.end3 = Location( self, address=1, type_ = "end3" )
891
892
        for l in (self.start5,self.end3):
            self.locations.append(l)
893

894
    def connect_end3(self, end5, force_connection=False):
895
896
        self._connect_end( end5,  _5_to_3 = False, force_connection = force_connection )

897
    def connect_5end(self, end3, force_connection=False): # TODO: change name or possibly deprecate
898
899
900
901
902
903
904
        self._connect_end( end3,  _5_to_3 = True, force_connection = force_connection )

    def _connect_end(self, other, _5_to_3, force_connection):
        assert( isinstance(other, Location) )
        if _5_to_3 == True:
            my_end = self.end5
            assert( other.type_ == "end3" )
905
906
            conn = Connection( my_end, other, type_="intrahelical" )
            self._connect( other.container, conn, in_3prime_direction=True )
907
908
909
        else:
            my_end = self.end3
            assert( other.type_ == "end5" )
910
911
            conn = Connection( other, my_end, type_="intrahelical" )
            other.container._connect( self, conn, in_3prime_direction=True )
912

913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
    def add_crossover(self, nt, other, other_nt, strands_fwd=[True,False]):
        """ Add a crossover between two helices """
        ## Validate other, nt, other_nt
        ##   TODO
       
        ## TODO: fix direction

        c1 = self.nt_pos_to_contour(nt)
        ## Ensure connections occur at ends, otherwise the structure doesn't make sense
        assert(np.isclose(c1,0) or np.isclose(c1,1))
        loc = self.get_location_at(c1, True)

        c2 = other.nt_pos_to_contour(other_nt)
        if isinstance(other,SingleStrandedSegment):
            ## Ensure connections occur at opposing ends
            assert(np.isclose(c2,0) or np.isclose(c2,1))
            other_loc = other.get_location_at( c2, True )
            assert( loc.type_ in ("end3","end5"))
            assert( other_loc.type_ in ("end3","end5"))
            if loc.type_ == "end3":
                self.connect_end3( other_loc )
            else:
                assert( other_loc.type_ == "end3" )
                other.connect_end3( self )

        else:
            assert( loc.type_ in ("end3","end5"))
            assert(c2 >= 0 and c2 <= 1)
            other_loc = other.get_location_at( c2, strands_fwd[1] )
            if loc.type_ == "end3":
                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 )
946

947
948
    def _get_num_beads(self, contour, max_basepairs_per_bead, max_nucleotides_per_bead):
        return int(contour*self.num_nts // max_nucleotides_per_bead)
cmaffeo2's avatar
cmaffeo2 committed
949

950
951
    def _generate_one_bead(self, contour_position, nts):
        pos = self.contour_to_position(contour_position)
952
953
        b = SegmentParticle( Segment.ssDNA_particle, pos, 
                             name="NAS",
954
955
956
957
                             num_nts=nts, parent=self,
                             contour_position=contour_position )
        self._add_bead(b)
        return b
958
959

    
cmaffeo2's avatar
cmaffeo2 committed
960
961
962
963
class StrandInSegment(Group):
    """ Class that holds atomic model, maps to segment """
    
    def __init__(self, segment, start, end, is_fwd):
964
        """ start/end should be provided expressed in nt coordinates, is_fwd tuples """
cmaffeo2's avatar
cmaffeo2 committed
965
966
        Group.__init__(self)
        self.num_nts = 0
967
        # self.sequence = []
cmaffeo2's avatar
cmaffeo2 committed
968
969
970
971
972
        self.segment = segment
        self.start = start
        self.end = end
        self.is_fwd = is_fwd

973
        nts = np.abs(end-start)+1
974
        self.num_nts = int(round(nts))
975
        assert( np.abs(self.num_nts-nts) < 1e-5 )
cmaffeo2's avatar
cmaffeo2 committed
976

977
        # print(" Creating {}-nt StrandInSegment in {} from {} to {} {}".format(self.num_nts, segment.name, start, end, is_fwd))
978
979
980
981
982
983
984
985
986
    
    def _nucleotide_ids(self):
        nt0 = self.start # seg.contour_to_nt_pos(self.start)
        assert( np.abs(nt0 - round(nt0)) < 1e-5 )
        nt0 = int(round(nt0))
        assert( (self.end-self.start) >= 0 or not self.is_fwd )

        direction = (2*self.is_fwd-1)
        return range(nt0,nt0 + direction*self.num_nts, direction)
987
988
989

    def get_sequence(self):
        """ return 5-to-3 """
990
        # TODOTODO test
991
992
        seg = self.segment
        if self.is_fwd:
993
            return [seg.sequence[nt] for nt in self._nucleotide_ids()]
994
        else:
995
            return [seqComplement[seg.sequence[nt]] for nt in self._nucleotide_ids()]
996
997
998
999
1000
    
    def get_contour_points(self):
        c0,c1 = [self.segment.nt_pos_to_contour(p) for p in (self.start,self.end)]
        return np.linspace(c0,c1,self.num_nts)
            
For faster browsing, not all history is shown. View entire blame