segmentmodel.py 124 KB
Newer Older
cmaffeo2's avatar
cmaffeo2 committed
1
import pdb
2
from pathlib import Path
3
import numpy as np
4
import random
5
6
7
from .model.arbdmodel import PointParticle, ParticleType, Group, ArbdModel
from .coords import rotationAboutAxis, quaternion_from_matrix, quaternion_to_matrix
from .model.nonbonded import *
8
from copy import copy, deepcopy
9
from .model.nbPot import nbDnaScheme
10

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

183
    def _connect(self, other, connection, in_3prime_direction=None):
184
185
        ## TODO fix circular references        
        A,B = [connection.A, connection.B]
186
187
188
189
        if in_3prime_direction is not None:
            A.is_3prime_side_of_connection = not in_3prime_direction
            B.is_3prime_side_of_connection = in_3prime_direction
            
190
        A.connection = B.connection = connection
191
        self.connections.append(connection)
192
193
        if other is not self:
            other.connections.append(connection)
194
195
196
197
198
199
        l = A.container.locations
        if A not in l: l.append(A)
        l = B.container.locations
        if B not in l: l.append(B)
        

200
201
    # def _find_connections(self, loc):
    #     return [c for c in self.connections if c.A == loc or c.B == loc]
202
203

class SegmentParticle(PointParticle):
cmaffeo2's avatar
cmaffeo2 committed
204
    def __init__(self, type_, position, name="A", **kwargs):
205
        self.name = name
206
        self.contour_position = None
cmaffeo2's avatar
cmaffeo2 committed
207
        PointParticle.__init__(self, type_, position, name=name, **kwargs)
208
209
        self.intrahelical_neighbors = []
        self.other_neighbors = []
cmaffeo2's avatar
cmaffeo2 committed
210
        self.locations = []
211
212
213

    def get_intrahelical_above(self):
        """ Returns bead directly above self """
214
        # assert( len(self.intrahelical_neighbors) <= 2 )
215
216
217
218
219
220
        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 """
221
        # assert( len(self.intrahelical_neighbors) <= 2 )
222
223
224
225
        for b in self.intrahelical_neighbors:
            if b.get_contour_position(self.parent) < self.contour_position:
                return b

cmaffeo2's avatar
cmaffeo2 committed
226
    def _neighbor_should_be_added(self,b):
227
228
229
        if type(self.parent) != type(b.parent):
            return True

cmaffeo2's avatar
cmaffeo2 committed
230
231
232
233
        c1 = self.contour_position
        c2 = b.get_contour_position(self.parent)
        if c2 < c1:
            b0 = self.get_intrahelical_below()
234
        else:
cmaffeo2's avatar
cmaffeo2 committed
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
            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:
252
253
            # assert(len(b.intrahelical_neighbors) <= 1)
            # assert(len(self.intrahelical_neighbors) <= 1)
cmaffeo2's avatar
cmaffeo2 committed
254
255
            self.intrahelical_neighbors.append(b)
            b.intrahelical_neighbors.append(self)
256

cmaffeo2's avatar
cmaffeo2 committed
257
258
259
260
261
262
263
264
265
266
    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)

267
268
269
270
    def get_contour_position(self,seg):
        if seg == self.parent:
            return self.contour_position
        else:
cmaffeo2's avatar
cmaffeo2 committed
271
272
273
274
275
            cutoff = 30*3
            target_seg = seg

            ## depth-first search
            ## TODO cache distances to nearby locations?
cmaffeo2's avatar
cmaffeo2 committed
276
            def descend_search_tree(seg, contour_in_seg, distance=0, visited_segs=None):
cmaffeo2's avatar
cmaffeo2 committed
277
                nonlocal cutoff
cmaffeo2's avatar
cmaffeo2 committed
278
                if visited_segs is None: visited_segs = []
cmaffeo2's avatar
cmaffeo2 committed
279
280
281
282

                if seg == target_seg:
                    # pdb.set_trace()
                    ## Found a segment in our target
283
284
                    sign = 1 if contour_in_seg == 1 else -1
                    if sign == -1: assert( contour_in_seg == 0 )
cmaffeo2's avatar
cmaffeo2 committed
285
286
                    if distance < cutoff: # TODO: check if this does anything
                        cutoff = distance
287
                    return [[distance, contour_in_seg+sign*seg.nt_pos_to_contour(distance)]], [(seg, contour_in_seg, distance)]
cmaffeo2's avatar
cmaffeo2 committed
288
                if distance > cutoff:
289
                    return None,None
cmaffeo2's avatar
cmaffeo2 committed
290
291
                    
                ret_list = []
292
                hist_list = []
cmaffeo2's avatar
cmaffeo2 committed
293
                ## Find intrahelical locations in seg that we might pass through
294
295
296
297
298
                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
299
                    if B.container in visited_segs: continue
300
301
302
                    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
303
304
305
                                                   distance+dx, visited_segs + [seg] )
                    if results is not None:
                        ret_list.extend( results )
306
307
                        hist_list.extend( history )
                return ret_list,hist_list
cmaffeo2's avatar
cmaffeo2 committed
308

309
            results,history = descend_search_tree(self.parent, self.contour_position)
cmaffeo2's avatar
cmaffeo2 committed
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
            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
329
330
331
    def __repr__(self):
        return "<SegmentParticle {} on {}[{:.2f}]>".format( self.name, self.parent, self.contour_position)

332
333

## TODO break this class into smaller, better encapsulated pieces
334
335
336
337
338
339
340
341
342
343
344
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
345
346
347
348
349
    orientation_particle = ParticleType("O",
                                        diffusivity = 100,
                                        mass = 300,
                                        radius = 1,
                                    )
350

cmaffeo2's avatar
cmaffeo2 committed
351
    # orientation_bond = HarmonicBond(10,2)
352
    orientation_bond = HarmonicBond(30,1.5, rRange = (0,500) )
353
354
355
356
357
358
359

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

360
    def __init__(self, name, num_nt, 
cmaffeo2's avatar
cmaffeo2 committed
361
                 start_position = None,
362
                 end_position = None, 
cmaffeo2's avatar
cmaffeo2 committed
363
364
                 segment_model = None,
                 **kwargs):
365

cmaffeo2's avatar
cmaffeo2 committed
366
367
        if start_position is None: start_position = np.array((0,0,0))

cmaffeo2's avatar
cmaffeo2 committed
368
        Group.__init__(self, name, children=[], **kwargs)
369
        ConnectableElement.__init__(self, connection_locations=[], connections=[])
370

cmaffeo2's avatar
cmaffeo2 committed
371
372
373
        if 'segname' not in kwargs:
            self.segname = name
        # self.resname = name
cmaffeo2's avatar
cmaffeo2 committed
374
375
376
377
378
        self.start_orientation = None
        self.twist_per_nt = 0

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

379
380
381
        self._bead_model_generation = 0    # TODO: remove?
        self.segment_model = segment_model # TODO: remove?

382
383
384
385
        self.strand_pieces = dict()
        for d in ('fwd','rev'):
            self.strand_pieces[d] = []

386
        self.num_nt = int(num_nt)
387
        if end_position is None:
388
            end_position = np.array((0,0,self.distance_per_nt*num_nt)) + start_position
389
390
391
        self.start_position = start_position
        self.end_position = end_position

cmaffeo2's avatar
cmaffeo2 committed
392
393
394
395
        ## Used to assign cadnano names to beads
        self._generate_bead_callbacks = []
        self._generate_nucleotide_callbacks = []

396
        ## Set up interpolation for positions
397
398
399
400
        self._set_splines_from_ends()

        self.sequence = None

401
402
403
    def __repr__(self):
        return "<{} {}[{:d}]>".format( type(self), self.name, self.num_nt )

cmaffeo2's avatar
cmaffeo2 committed
404
    def set_splines(self, contours, coords):
405
        tck, u = interpolate.splprep( coords.T, u=contours, s=0, k=1)
406
        self.position_spline_params = (tck,u)
407

408
409
    def set_orientation_splines(self, contours, quaternions):
        tck, u = interpolate.splprep( quaternions.T, u=contours, s=0, k=1)
410
411
412
413
414
415
        self.quaternion_spline_params = (tck,u)

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

416
417
418
    def _get_location_positions(self):
        return [self.contour_to_nt_pos(l.address) for l in self.locations]

cmaffeo2's avatar
cmaffeo2 committed
419
    def insert_dna(self, at_nt: int, num_nt: int, seq=tuple()):
420
421
422
423
424
425
426
427
428
429
430
431
        assert(np.isclose(np.around(num_nt),num_nt))
        if at_nt < 0:
            raise ValueError("Attempted to insert DNA into {} at a negative location".format(self))
        if at_nt > self.num_nt-1:
            raise ValueError("Attempted to insert DNA into {} at beyond the end of the Segment".format(self))
        if num_nt < 0:
            raise ValueError("Attempted to insert DNA a negative amount of DNA into {}".format(self))

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

cmaffeo2's avatar
cmaffeo2 committed
432
433
        ## TODO: handle sequence

434
435
436
437
438
439
        self.num_nt = self.num_nt+num_nt

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

    def remove_dna(self, first_nt: int, last_nt: int):
cmaffeo2's avatar
cmaffeo2 committed
440
        """ Removes nucleotides between first_nt and last_nt, inclusive """
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
        assert(np.isclose(np.around(first_nt),first_nt))
        assert(np.isclose(np.around(last_nt),last_nt))
        tmp = min((first_nt,last_nt))
        last_nt = max((first_nt,last_nt))
        fist_nt = tmp

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

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

        nt_positions = self._get_location_positions()

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

cmaffeo2's avatar
cmaffeo2 committed
463
464
465
466
467
468
469
470
        removed_nt = last_nt-first_nt+1
        new_nt_positions = [p if p <= last_nt else p-removed_nt for p in nt_positions]
        num_nt = self.num_nt-removed_nt

        if self.sequence is not None and len(self.sequence) == self.num_nt:
            self.sequence = [s for s,i in zip(self.sequence,range(self.num_nt)) 
                                if i < first_nt or i > last_nt]
            assert( len(self.sequence) == num_nt )
471

cmaffeo2's avatar
cmaffeo2 committed
472
        self.num_nt = num_nt
473
474
475
476

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

477
478
479
480
481
482
483
484
485
486
487
488
    def __filter_contours(contours, positions, position_filter, contour_filter):
        u = contours
        r = positions

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

489
    def translate(self, translation_vector, position_filter=None, contour_filter=None):
490
491
492
493
494
495
496
497
        dr = np.array(translation_vector)
        tck, u = self.position_spline_params
        r = self.contour_to_position(u)

        ids = Segment.__filter_contours(u, r, position_filter, contour_filter)

        ## Translate
        r[ids,:] = r[ids,:] + dr[np.newaxis,:]
498
        self.set_splines(u,r)
499
500
501
502
503
504
505
506
507

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

        ids = Segment.__filter_contours(u, r, position_filter, contour_filter)

        if about is None:
            ## TODO: do this more efficiently
508
            r[ids,:] = np.array([rotation_matrix.dot(r[i,:]) for i in ids])
509
510
511
        else:
            dr = np.array(about)
            ## TODO: do this more efficiently
512
            r[ids,:] = np.array([rotation_matrix.dot(r[i,:]-dr) + dr for i in ids])
513

514
        self.set_splines(u,r)
515
516
517
518

        if self.quaternion_spline_params is not None:
            ## TODO: performance: don't shift between quaternion and matrix representations so much
            tck, u = self.quaternion_spline_params
519
            orientations = [self.contour_to_orientation(v) for v in u]
520
521
522
            for i in ids:
                orientations[i,:] = rotation_matrix.dot(orientations[i])
            quats = [quaternion_from_matrix(o) for o in orientations]
523
            self.set_orientation_splines(u, quats)
524

525
    def _set_splines_from_ends(self, resolution=4):
526
        self.quaternion_spline_params = None
527
528
529
530
531
532
        r0 = np.array(self.start_position)[np.newaxis,:]
        r1 = np.array(self.end_position)[np.newaxis,:]
        u = np.linspace(0,1, max(3,self.num_nt//int(resolution)))
        s = u[:,np.newaxis]
        coords = (1-s)*r0 + s*r1
        self.set_splines(u, coords)
533

534
535
536
    def clear_all(self):
        Group.clear_all(self)  # TODO: use super?
        self.beads = []
537
538
539
540
541
        # for c,loc,other in self.get_connections_and_locations():
        #     loc.particle = None
        #     other.particle = None
        for l in self.locations:
            l.particle = None
542

543
    def contour_to_nt_pos(self, contour_pos, round_nt=False):
544
        nt = contour_pos*(self.num_nt) - 0.5
545
        if round_nt:
cmaffeo2's avatar
cmaffeo2 committed
546
            assert( np.isclose(np.around(nt),nt) )
547
548
549
            nt = np.around(nt)
        return nt

550
    def nt_pos_to_contour(self,nt_pos):
551
        return (nt_pos+0.5)/(self.num_nt)
552

553
    def contour_to_position(self,s):
554
        p = interpolate.splev( s, self.position_spline_params[0] )
555
556
557
558
        if len(p) > 1: p = np.array(p).T
        return p

    def contour_to_tangent(self,s):
559
        t = interpolate.splev( s, self.position_spline_params[0], der=1 )
560
561
        t = (t / np.linalg.norm(t,axis=0))
        return t.T
562
563
564
        

    def contour_to_orientation(self,s):
565
        assert( isinstance(s,float) or isinstance(s,int) or len(s) == 1 )   # TODO make vectorized version
566

567
568
569
570
571
        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)
572
            zAxis = np.array((0,0,1))
573
574

            if rotAxisL > 0.001:
575
576
577
                theta = np.arcsin(rotAxisL) * 180/np.pi
                if axis.dot(zAxis) < 0: theta = 180-theta
                orientation0 = rotationAboutAxis( rotAxis/rotAxisL, theta, normalizeAxis=False ).T
578
            else:
579
580
                orientation0 = np.eye(3) if axis.dot(zAxis) > 0 else \
                               rotationAboutAxis( np.array((1,0,0)), 180, normalizeAxis=False )
581
582
583
            if self.start_orientation is not None:
                orientation0 = orientation0.dot(self.start_orientation)

584
585
586
            orientation = rotationAboutAxis( axis, self.twist_per_nt*self.contour_to_nt_pos(s), normalizeAxis=False )
            orientation = orientation.dot(orientation0)
        else:
587
            q = interpolate.splev( s, self.quaternion_spline_params[0] )
588
589
            if len(q) > 1: q = np.array(q).T # TODO: is this needed?
            orientation = quaternion_to_matrix(q)
590

591
        return orientation
592

cmaffeo2's avatar
cmaffeo2 committed
593
    def get_contour_sorted_connections_and_locations(self,type_):
cmaffeo2's avatar
cmaffeo2 committed
594
        sort_fn = lambda c: c[1].address
cmaffeo2's avatar
cmaffeo2 committed
595
        cl = self.get_connections_and_locations(type_)
cmaffeo2's avatar
cmaffeo2 committed
596
        return sorted(cl, key=sort_fn)
597
598
599
    
    def randomize_unset_sequence(self):
        bases = list(seqComplement.keys())
600
        # bases = ['T']        ## FOR DEBUG
601
        if self.sequence is None:
602
            self.sequence = [random.choice(bases) for i in range(self.num_nt)]
603
        else:
604
            assert(len(self.sequence) == self.num_nt) # TODO move
605
606
607
            for i in range(len(self.sequence)):
                if self.sequence[i] is None:
                    self.sequence[i] = random.choice(bases)
608

cmaffeo2's avatar
cmaffeo2 committed
609
610
611
    def _get_num_beads(self, max_basepairs_per_bead, max_nucleotides_per_bead ):
        raise NotImplementedError

612
    def _generate_one_bead(self, contour_position, nts):
613
614
        raise NotImplementedError

cmaffeo2's avatar
cmaffeo2 committed
615
    def _generate_atomic_nucleotide(self, contour_position, is_fwd, seq, scale, strand_segment):
cmaffeo2's avatar
cmaffeo2 committed
616
617
618
619
620
        """ 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)
621
        orientation = self.contour_to_orientation(contour_position)
cmaffeo2's avatar
cmaffeo2 committed
622

623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
        """ 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
641
        key = seq
642
643
        nt_dict = canonicalNtFwd if is_fwd else canonicalNtRev

644
        atoms = nt_dict[ key ].generate() # TODO: clone?
cmaffeo2's avatar
cmaffeo2 committed
645
        atoms.orientation = orientation.dot(atoms.orientation)
646
647
648
649
        if isinstance(self, SingleStrandedSegment):
            if scale is not None and scale != 1:
                for a in atoms:
                    a.position = scale*a.position
650
            atoms.position = pos - atoms.atoms_by_name["C1'"].collapsedPosition()
651
652
653
654
655
656
657
658
659
        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
660
                    else:
661
                        a.fixed = 1
662
            atoms.position = pos
cmaffeo2's avatar
cmaffeo2 committed
663
664
665
666
667
668
        
        atoms.contour_position = contour_position
        strand_segment.add(atoms)

        for callback in self._generate_nucleotide_callbacks:
            callback(atoms)
cmaffeo2's avatar
cmaffeo2 committed
669
670

        return atoms
671

672
673
    def add_location(self, nt, type_, on_fwd_strand=True):
        ## Create location if needed, add to segment
674
        c = self.nt_pos_to_contour(nt)
675
676
677
678
679
680
        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?
681
682
683
684
685

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

686
    def add_5prime(self, nt, on_fwd_strand=True):
687
688
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
689
        self.add_location(nt,"5prime",on_fwd_strand)
690
691

    def add_3prime(self, nt, on_fwd_strand=True):
692
693
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
694
        self.add_location(nt,"3prime",on_fwd_strand)
695

696
    def get_3prime_locations(self):
cmaffeo2's avatar
cmaffeo2 committed
697
        return sorted(self.get_locations("3prime"),key=lambda x: x.address)
698
    
cmaffeo2's avatar
cmaffeo2 committed
699
    def get_5prime_locations(self):
700
        ## TODO? ensure that data is consistent before _build_model calls
cmaffeo2's avatar
cmaffeo2 committed
701
        return sorted(self.get_locations("5prime"),key=lambda x: x.address)
cmaffeo2's avatar
cmaffeo2 committed
702

703
    def iterate_connections_and_locations(self, reverse=False):
cmaffeo2's avatar
cmaffeo2 committed
704
705
        ## connections to other segments
        cl = self.get_contour_sorted_connections_and_locations()
706
        if reverse:
cmaffeo2's avatar
cmaffeo2 committed
707
            cl = cl[::-1]
708
709
710
            
        for c in cl:
            yield c
cmaffeo2's avatar
cmaffeo2 committed
711

712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
    ## 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)

730
    ## TODO rename
731
    def get_strand_segment(self, nt_pos, is_fwd, move_at_least=0.5):
732
        """ Walks through locations, checking for crossovers """
733
734
735
736
        # if self.name in ("6-1","1-1"):
        #     import pdb
        #     pdb.set_trace()
        move_at_least = 0
737
738

        ## Iterate through locations
cmaffeo2's avatar
cmaffeo2 committed
739
        # locations = sorted(self.locations, key=lambda l:(l.address,not l.on_fwd_strand), reverse=(not is_fwd))
740
741
742
743
744
745
        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))
746
747
        # print(locations)

748
        for l in locations:
cmaffeo2's avatar
cmaffeo2 committed
749
750
751
752
            # TODOTODO probably okay
            if l.address == 0:
                pos = 0.0
            elif l.address == 1:
753
                pos = self.num_nt-1
cmaffeo2's avatar
cmaffeo2 committed
754
755
            else:
                pos = self.contour_to_nt_pos(l.address, round_nt=True)
756
757
758

            ## DEBUG

cmaffeo2's avatar
cmaffeo2 committed
759

760
            ## Skip locations encountered before our strand
761
762
763
764
765
766
767
768
            # 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
769
770

            ## Stop if we found the 3prime end
771
            if l.on_fwd_strand == is_fwd and l.type_ == "3prime" and l.connection is None:
772
                # print("  found end at",l)
773
                return pos, None, None, None, None
774
775
776
777
778
779
780
781

            ## 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:
782
783
                # print("  passing through",l)
                # print("from {}, connection {} to {}".format(nt_pos,l,B))
784
                Bpos = B.get_nt_pos()
785
                return pos, B.container, Bpos, B.on_fwd_strand, 0.5
786
787
788
                
            ## Stop at other strand crossovers so basepairs line up
            elif c.type_ == "crossover":
789
                if nt_pos == pos: continue
790
                # print("  pausing at",l)
791
                return pos, l.container, pos+(2*is_fwd-1), is_fwd, 0
792
793
794
795
796
797

        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

798
799
800
    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
801
        # TODO: include beads in connections?
802
803
804
        i = np.argmin((cs - contour_position)**2)

        return self.beads[i]
805

806
807
808
809
810
811
812
813
814
    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 ))

815
816
    def get_all_consecutive_beads(self, number):
        assert(number >= 1)
cmaffeo2's avatar
cmaffeo2 committed
817
        ## Assume that consecutive beads in self.beads are bonded
818
        ret = []
cmaffeo2's avatar
cmaffeo2 committed
819
820
        for i in range(len(self.beads)-number+1):
            tmp = [self.beads[i+j] for j in range(0,number)]
821
            ret.append( tmp )
822
        return ret   
823

824
825
826
    def _add_bead(self,b,set_contour=False):
        if set_contour:
            b.contour_position = b.get_contour_position(self)
827
        
828
829
830
        # assert(b.parent is None)
        if b.parent is not None:
            b.parent.children.remove(b)
831
        self.add(b)
832
833
834
835
836
837
        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)
838
            self.add(o)
839
840
841
842
843
844
845
846
            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 = []
847
848

        if True:
849
850
            ## TODO: remove this if duplicates are never found 
            # print("Searching for duplicate particles...")
851
            ## Remove duplicates, preserving order
852
853
854
855
            tmp = []
            for c in new_children:
                if c not in tmp:
                    tmp.append(c)
856
                else:
857
                    print("  DUPLICATE PARTICLE FOUND!")
858
859
            new_children = tmp

860
861
862
863
864
        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)
865
866
867
868
869
            
        # 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)
870
871
        assert(len(old_children) == len(self.children))
        assert(len(old_beads) == len(self.beads))
872

873

cmaffeo2's avatar
cmaffeo2 committed
874
    def _generate_beads(self, bead_model, max_basepairs_per_bead, max_nucleotides_per_bead):
875

876
        """ Generate beads (positions, types, etc) and bonds, angles, dihedrals, exclusions """
cmaffeo2's avatar
cmaffeo2 committed
877
        ## TODO: decide whether to remove bead_model argument
878
        ##       (currently unused)
cmaffeo2's avatar
cmaffeo2 committed
879

880
        ## First find points between-which beads must be generated
881
882
883
        # 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
884
885
        # if self.name == "S001":
        #     pdb.set_trace()
886

cmaffeo2's avatar
cmaffeo2 committed
887
888
        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) )
889

890
891
892
        # if self.num_nt == 1 and all([l.particle is not None for l in self.locations]):
        #     pdb.set_trace()
        #     return
893

894
895
896
897
        for b in existing_beads:
            assert(b.parent is not None)

        ## Add ends if they don't exist yet
898
        ## TODOTODO: test 1 nt segments?
899
        if len(existing_beads) == 0 or existing_beads[0].get_nt_position(self) >= 0.5:
cmaffeo2's avatar
cmaffeo2 committed
900
901
            # if len(existing_beads) > 0:            
            #     assert(existing_beads[0].get_nt_position(self) >= 0.5)
902
            b = self._generate_one_bead( self.nt_pos_to_contour(0), 0)
903
            existing_beads = [b] + existing_beads
cmaffeo2's avatar
cmaffeo2 committed
904

cmaffeo2's avatar
cmaffeo2 committed
905
        if existing_beads[-1].get_nt_position(self)-(self.num_nt-1) < -0.5 or len(existing_beads)==1:
906
            b = self._generate_one_bead( self.nt_pos_to_contour(self.num_nt-1), 0)
907
908
909
910
911
            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
912
        last = None
cmaffeo2's avatar
cmaffeo2 committed
913

914
915
        for I in range(len(existing_beads)-1):
            eb1,eb2 = [existing_beads[i] for i in (I,I+1)]
916
            assert( eb1 is not eb2 )
917

cmaffeo2's avatar
cmaffeo2 committed
918
919
920
921
            # if np.isclose(eb1.position[2], eb2.position[2]):
            #     import pdb
            #     pdb.set_trace()

922
            # print(" %s working on %d to %d" % (self.name, eb1.position[2], eb2.position[2]))
923
924
            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
925
926
927
928
929
930

            ## 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

931
            ds = e_ds / (num_beads+1)
932
933
934
            nts = ds*self.num_nt
            eb1.num_nt += 0.5*nts
            eb2.num_nt += 0.5*nts
935
936
937
938
939
940
941

            ## 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
942
                last.make_intrahelical_neighbor(eb1)
943
944
945
            last = eb1
            for j in range(num_beads):
                s = ds*(j+1) + s0
946
                # if self.name in ("51-2","51-3"):
cmaffeo2's avatar
cmaffeo2 committed
947
                # if self.name in ("31-2",):
948
                #     print(" adding bead at {}".format(s))
949
950
                b = self._generate_one_bead(s,nts)

cmaffeo2's avatar
cmaffeo2 committed
951
                last.make_intrahelical_neighbor(b)
952
953
954
                last = b
                tmp_children.append(b)

cmaffeo2's avatar
cmaffeo2 committed
955
        last.make_intrahelical_neighbor(eb2)
956
957
958

        if eb2.parent == self:
            tmp_children.append(eb2)
cmaffeo2's avatar
cmaffeo2 committed
959
        # if self.name in ("31-2",):
960
        #     pdb.set_trace()
961
        self._rebuild_children(tmp_children)
962

cmaffeo2's avatar
cmaffeo2 committed
963
964
965
        for callback in self._generate_bead_callbacks:
            callback(self)

966
967
968
969
970
971
972
973
974
    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 """

975
    def __init__(self, name, num_bp, start_position = np.array((0,0,0)),
976
977
                 end_position = None, 
                 segment_model = None,
cmaffeo2's avatar
cmaffeo2 committed
978
979
                 local_twist = False,
                 num_turns = None,