segmentmodel.py 163 KB
Newer Older
cmaffeo2's avatar
cmaffeo2 committed
1
import pdb
2
import subprocess
3
from pathlib import Path
4
import numpy as np
5
import random
6
7
8
from .arbdmodel import PointParticle, ParticleType, Group, ArbdModel
from .arbdmodel.coords import rotationAboutAxis, quaternion_from_matrix, quaternion_to_matrix
from .arbdmodel.nonbonded import *
9
from copy import copy, deepcopy
10
from .model.nbPot import nbDnaScheme
11

12
13
import re

cmaffeo2's avatar
cmaffeo2 committed
14
15
from scipy.special import erf
import scipy.optimize as opt
16
from scipy import interpolate
cmaffeo2's avatar
cmaffeo2 committed
17

18
19
from .model.CanonicalNucleotideAtoms import canonicalNtFwd, canonicalNtRev, seqComplement
from .model.CanonicalNucleotideAtoms import enmTemplateHC, enmTemplateSQ, enmCorrectionsHC
cmaffeo2's avatar
cmaffeo2 committed
20

cmaffeo2's avatar
cmaffeo2 committed
21
from .model.spring_from_lp import k_angle as angle_spring_from_lp
22
from .model.spring_from_lp import k_twist as twist_spring_from_lp
cmaffeo2's avatar
cmaffeo2 committed
23

24
25
import csv

26
# import pdb
27
"""
cmaffeo2's avatar
cmaffeo2 committed
28
TODO:
cmaffeo2's avatar
cmaffeo2 committed
29
 + fix handling of crossovers for atomic representation
cmaffeo2's avatar
cmaffeo2 committed
30
 + map to atomic representation
31
    + add nicks
cmaffeo2's avatar
cmaffeo2 committed
32
    + transform ssDNA nucleotides 
cmaffeo2's avatar
cmaffeo2 committed
33
    - shrink ssDNA
cmaffeo2's avatar
cmaffeo2 committed
34
    + shrink dsDNA backbone
35
    + make orientation continuous
cmaffeo2's avatar
cmaffeo2 committed
36
    + sequence
37
    + handle circular dna
38
 + ensure crossover bead potentials aren't applied twice 
39
 + remove performance bottlenecks
40
 - test for large systems
cmaffeo2's avatar
cmaffeo2 committed
41
 + assign sequence
42
 + ENM
43
44
 - rework Location class 
 - remove recursive calls
45
 - document
46
 - develop unit test suite
47
48
 - refactor parts of Segment into an abstract_polymer class
 - make each call generate_bead_model, generate_atomic_model, generate_oxdna_model return an object with only have a reference to original object
49
"""
50
51
52

_DEBUG_TRACE = False

cmaffeo2's avatar
cmaffeo2 committed
53
54
class CircularDnaError(Exception):
    pass
55

cmaffeo2's avatar
cmaffeo2 committed
56
57
58
class ParticleNotConnectedError(Exception):
    pass

59
60
class Location():
    """ Site for connection within an object """
61
    def __init__(self, container, address, type_, on_fwd_strand = True):
62
        ## TODO: remove cyclic references(?)
63
        self.container = container
cmaffeo2's avatar
cmaffeo2 committed
64
        self.address = address  # represents position along contour length in segment
cmaffeo2's avatar
cmaffeo2 committed
65
        # assert( type_ in ("end3","end5") ) # TODO remove or make conditional
66
        self.on_fwd_strand = on_fwd_strand
67
68
        self.type_ = type_
        self.particle = None
69
        self.connection = None
70
        self.is_3prime_side_of_connection = None
71

72
        self.combine = None     # some locations might be combined in bead model 
73
74
75
76
77
78
79

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

80
    def set_connection(self, connection, is_3prime_side_of_connection):
81
        self.connection = connection # TODO weakref? 
82
        self.is_3prime_side_of_connection = is_3prime_side_of_connection
83

84
85
86
87
88
89
90
    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:
91
                pos = self.container.num_nt-1
92
93
94
95
            else:
                raise
        return pos

96
97
98
99
100
101
    def delete(self):
        if self.connection is not None:
            self.connection.delete()
        if self.container is not None:
            self.container.locations.remove(self)

102
103
104
105
106
    def __repr__(self):
        if self.on_fwd_strand:
            on_fwd = "on_fwd_strand"
        else:
            on_fwd = "on_rev_strand"
107
108
109
110
        try:
            return "<Location {}.{}[{:d},{}]>".format( self.container.name, self.type_, self.get_nt_pos(), on_fwd )
        except:
            return "<Location {}.{}[{:.2f},{:d}]>".format( self.container.name, self.type_, self.address, self.on_fwd_strand)
111
        
112
113
114
115
116
117
118
119
120
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_
        
121
122
123
124
125
126
127
    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
128

129
130
    def delete(self):
        self.A.container.connections.remove(self)
131
132
        if self.B.container is not self.A.container:
            self.B.container.connections.remove(self)
133
134
135
        self.A.connection = None
        self.B.connection = None

cmaffeo2's avatar
cmaffeo2 committed
136
137
138
139
    def __repr__(self):
        return "<Connection {}--{}--{}]>".format( self.A, self.type_, self.B )
        

140
        
141
142
143
# class ConnectableElement(Transformable):
class ConnectableElement():
    """ Abstract base class """
cmaffeo2's avatar
cmaffeo2 committed
144
145
146
147
    def __init__(self, connection_locations=None, connections=None):
        if connection_locations is None: connection_locations = []
        if connections is None: connections = []

148
149
        ## TODO decide on names
        self.locations = self.connection_locations = connection_locations
150
151
        self.connections = connections

cmaffeo2's avatar
cmaffeo2 committed
152
    def get_locations(self, type_=None, exclude=()):
153
154
155
156
157
158
159
160
        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] ) )
161
162
        return locs

163
    def get_location_at(self, address, on_fwd_strand=True, new_type=None):
164
        loc = None
165
        if (self.num_nt == 1):
166
167
168
169
170
171
172
            # 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
173
            # assert( loc is not None )
174
175
176
177
178
        else:
            for l in self.locations:
                if l.address == address and l.on_fwd_strand == on_fwd_strand:
                    assert(loc is None)
                    loc = l
179
        if loc is None and new_type is not None:
180
181
            loc = Location( self, address=address, type_=new_type, on_fwd_strand=on_fwd_strand )
        return loc
182

cmaffeo2's avatar
cmaffeo2 committed
183
    def get_connections_and_locations(self, connection_type=None, exclude=()):
184
185
        """ Returns a list with each entry of the form:
            connection, location_in_self, location_in_other """
186
        type_ = connection_type
187
188
        ret = []
        for c in self.connections:
189
            if (type_ is None or c.type_ == type_) and c.type_ not in exclude:
190
                if   c.A.container is self:
191
                    ret.append( [c, c.A, c.B] )
192
                elif c.B.container is self:
193
194
                    ret.append( [c, c.B, c.A] )
                else:
195
196
                    import pdb
                    pdb.set_trace()
197
198
199
                    raise Exception("Object contains connection that fails to refer to object")
        return ret

200
    def _connect(self, other, connection, in_3prime_direction=None):
201
202
        ## TODO fix circular references        
        A,B = [connection.A, connection.B]
203
204
205
206
        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
            
207
        A.connection = B.connection = connection
208
        self.connections.append(connection)
209
210
        if other is not self:
            other.connections.append(connection)
211

212
213
214
215
216
217
        l = A.container.locations
        if A not in l: l.append(A)
        l = B.container.locations
        if B not in l: l.append(B)
        

218
219
    # def _find_connections(self, loc):
    #     return [c for c in self.connections if c.A == loc or c.B == loc]
220
221

class SegmentParticle(PointParticle):
cmaffeo2's avatar
cmaffeo2 committed
222
    def __init__(self, type_, position, name="A", **kwargs):
223
        self.name = name
224
        self.contour_position = None
cmaffeo2's avatar
cmaffeo2 committed
225
        PointParticle.__init__(self, type_, position, name=name, **kwargs)
226
227
        self.intrahelical_neighbors = []
        self.other_neighbors = []
cmaffeo2's avatar
cmaffeo2 committed
228
        self.locations = []
229

230
    def get_intrahelical_above(self, all_types=True):
231
        """ Returns bead directly above self """
232
        # assert( len(self.intrahelical_neighbors) <= 2 )
233
        for b in self.intrahelical_neighbors:
234
            if b.get_contour_position(self.parent, self.contour_position) > self.contour_position:
235
236
                if all_types or isinstance(b,type(self)):
                    return b
237

238
    def get_intrahelical_below(self, all_types=True):
239
        """ Returns bead directly below self """
240
        # assert( len(self.intrahelical_neighbors) <= 2 )
241
        for b in self.intrahelical_neighbors:
242
            if b.get_contour_position(self.parent, self.contour_position) < self.contour_position:
243
244
                if all_types or isinstance(b,type(self)):
                    return b
245

cmaffeo2's avatar
cmaffeo2 committed
246
    def _neighbor_should_be_added(self,b):
247
248
249
        if type(self.parent) != type(b.parent):
            return True

cmaffeo2's avatar
cmaffeo2 committed
250
        c1 = self.contour_position
251
        c2 = b.get_contour_position(self.parent,c1)
cmaffeo2's avatar
cmaffeo2 committed
252
253
        if c2 < c1:
            b0 = self.get_intrahelical_below()
254
        else:
cmaffeo2's avatar
cmaffeo2 committed
255
256
257
            b0 = self.get_intrahelical_above()

        if b0 is not None:            
258
            c0 = b0.get_contour_position(self.parent,c1)
cmaffeo2's avatar
cmaffeo2 committed
259
260
261
262
263
264
265
266
267
268
269
270
271
            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:
272
273
            # assert(len(b.intrahelical_neighbors) <= 1)
            # assert(len(self.intrahelical_neighbors) <= 1)
cmaffeo2's avatar
cmaffeo2 committed
274
275
            self.intrahelical_neighbors.append(b)
            b.intrahelical_neighbors.append(self)
276

277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
    def conceptual_get_position(self, context):
        """ 
        context: object

Q: does this function do too much?
Danger of mixing return values

Q: does context describe the system or present an argument?
        """

        ## Validate Inputs
        ...

        ## Algorithm
        """
context specifies:
  - kind of output: real space, nt within segment, fraction of segment
  - absolute or relative
  - constraints: e.g. if passing through
        """
        """
given context, provide the position
        input
"""

    def get_nt_position(self, seg, near_address=None):
cmaffeo2's avatar
cmaffeo2 committed
303
304
305
306
        """ Returns the "address" of the nucleotide relative to seg in
        nucleotides, taking the shortest (intrahelical) contour length route to seg
        """
        if seg == self.parent:
307
            pos = self.contour_position
cmaffeo2's avatar
cmaffeo2 committed
308
        else:
309
310
            pos = self.get_contour_position(seg,near_address)
        return seg.contour_to_nt_pos(pos)
cmaffeo2's avatar
cmaffeo2 committed
311

312
    def get_contour_position(self,seg, address = None):
cmaffeo2's avatar
cmaffeo2 committed
313
314
315
316
        """ TODO: fix paradigm where a bead maps to exactly one location in a polymer
        - One way: modify get_contour_position to take an optional argument that indicates where in the polymer you are looking from
        """

317
318
319
        if seg == self.parent:
            return self.contour_position
        else:
cmaffeo2's avatar
cmaffeo2 committed
320
321
322
323
324
            cutoff = 30*3
            target_seg = seg

            ## depth-first search
            ## TODO cache distances to nearby locations?
cmaffeo2's avatar
cmaffeo2 committed
325
            def descend_search_tree(seg, contour_in_seg, distance=0, visited_segs=None):
cmaffeo2's avatar
cmaffeo2 committed
326
                nonlocal cutoff
cmaffeo2's avatar
cmaffeo2 committed
327
                if visited_segs is None: visited_segs = []
cmaffeo2's avatar
cmaffeo2 committed
328
329
330
331

                if seg == target_seg:
                    # pdb.set_trace()
                    ## Found a segment in our target
332
333
                    sign = 1 if contour_in_seg == 1 else -1
                    if sign == -1: assert( contour_in_seg == 0 )
cmaffeo2's avatar
cmaffeo2 committed
334
335
                    if distance < cutoff: # TODO: check if this does anything
                        cutoff = distance
336
                    return [[distance, contour_in_seg+sign*seg.nt_pos_to_contour(distance)]], [(seg, contour_in_seg, distance)]
cmaffeo2's avatar
cmaffeo2 committed
337
                if distance > cutoff:
338
                    return None,None
cmaffeo2's avatar
cmaffeo2 committed
339
340
                    
                ret_list = []
341
                hist_list = []
cmaffeo2's avatar
cmaffeo2 committed
342
                ## Find intrahelical locations in seg that we might pass through
343
344
345
346
347
                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
348
                    if B.container in visited_segs: continue
349
350
351
                    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
352
353
354
                                                   distance+dx, visited_segs + [seg] )
                    if results is not None:
                        ret_list.extend( results )
355
356
                        hist_list.extend( history )
                return ret_list,hist_list
cmaffeo2's avatar
cmaffeo2 committed
357

358
            results,history = descend_search_tree(self.parent, self.contour_position)
cmaffeo2's avatar
cmaffeo2 committed
359
360
            if results is None or len(results) == 0:
                raise Exception("Could not find location in segment") # TODO better error
361
362
363
364
            if address is not None:
                return sorted(results,key=lambda x:(x[0],(x[1]-address)**2))[0][1]
            else:
                return sorted(results,key=lambda x:x[0])[0][1]
cmaffeo2's avatar
cmaffeo2 committed
365
366
367
            # nt_pos = self.get_nt_position(seg)
            # return seg.nt_pos_to_contour(nt_pos)

368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
    def combine(self, other):
        assert(other in self.intrahelical_neighbors)
        assert(self in other.intrahelical_neighbors)

        self.intrahelical_neighbors.remove(other)
        other.intrahelical_neighbors.remove(self)
        for b in other.intrahelical_neighbors:
            b.intrahelical_neighbors.remove(other)
            b.intrahelical_neighbors.append(self)
            self.intrahelical_neighbors.append(b)
        for l in other.locations:
            self.locations.append(l)
            l.particle = self
        
        ## Remove bead
        other.parent.children.remove(other)
        if other in other.parent.beads:
            other.parent.beads.remove(other)
        if 'orientation_bead' in other.__dict__:
            other.parent.children.remove(other.orientation_bead)

        for b in list(other.parent.bonds):
            if other in b[:2]: other.parent.bonds.remove(b)

cmaffeo2's avatar
cmaffeo2 committed
392
393
394
395
396
397
398
399
400
401
402
403
    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
404
405
406
    def __repr__(self):
        return "<SegmentParticle {} on {}[{:.2f}]>".format( self.name, self.parent, self.contour_position)

407
408

## TODO break this class into smaller, better encapsulated pieces
409
410
411
412
413
414
415
416
417
418
419
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
420
421
422
423
424
    orientation_particle = ParticleType("O",
                                        diffusivity = 100,
                                        mass = 300,
                                        radius = 1,
                                    )
425

cmaffeo2's avatar
cmaffeo2 committed
426
    # orientation_bond = HarmonicBond(10,2)
427
    orientation_bond = HarmonicBond(30,1.5, rRange = (0,500) )
428
429
430
431
432
433
434

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

435
    def __init__(self, name, num_nt, 
cmaffeo2's avatar
cmaffeo2 committed
436
                 start_position = None,
437
                 end_position = None, 
cmaffeo2's avatar
cmaffeo2 committed
438
439
                 segment_model = None,
                 **kwargs):
440

cmaffeo2's avatar
cmaffeo2 committed
441
442
        if start_position is None: start_position = np.array((0,0,0))

cmaffeo2's avatar
cmaffeo2 committed
443
        Group.__init__(self, name, children=[], **kwargs)
444
        ConnectableElement.__init__(self, connection_locations=[], connections=[])
445

cmaffeo2's avatar
cmaffeo2 committed
446
447
448
        if 'segname' not in kwargs:
            self.segname = name
        # self.resname = name
cmaffeo2's avatar
cmaffeo2 committed
449
450
451
452
453
        self.start_orientation = None
        self.twist_per_nt = 0

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

454
455
456
        self._bead_model_generation = 0    # TODO: remove?
        self.segment_model = segment_model # TODO: remove?

457
458
459
460
        self.strand_pieces = dict()
        for d in ('fwd','rev'):
            self.strand_pieces[d] = []

461
        self.num_nt = int(num_nt)
462
        if end_position is None:
463
            end_position = np.array((0,0,self.distance_per_nt*num_nt)) + start_position
464
465
466
        self.start_position = start_position
        self.end_position = end_position

cmaffeo2's avatar
cmaffeo2 committed
467
468
469
470
        ## Used to assign cadnano names to beads
        self._generate_bead_callbacks = []
        self._generate_nucleotide_callbacks = []

471
        ## Set up interpolation for positions
472
473
474
475
        self._set_splines_from_ends()

        self.sequence = None

476
    def __repr__(self):
477
        return "<{} {}[{:d}]>".format( str(type(self)).split('.')[-1], self.name, self.num_nt )
478

cmaffeo2's avatar
cmaffeo2 committed
479
    def set_splines(self, contours, coords):
480
        tck, u = interpolate.splprep( coords.T, u=contours, s=0, k=1)
481
        self.position_spline_params = (tck,u)
482

483
484
    def set_orientation_splines(self, contours, quaternions):
        tck, u = interpolate.splprep( quaternions.T, u=contours, s=0, k=1)
485
486
487
488
489
490
        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)

491
492
493
494
    def get_bounding_box( self, num_points=3 ):
        positions = np.zeros( (num_points, 3) )
        i = 0
        for c in np.linspace(0,1,num_points):
495
            positions[i] = self.contour_to_position(c)
496
497
498
499
500
            i += 1
        min_ = np.array([np.min(positions[:,i]) for i in range(3)])
        max_ = np.array([np.max(positions[:,i]) for i in range(3)])
        return min_,max_

501
502
503
    def _get_location_positions(self):
        return [self.contour_to_nt_pos(l.address) for l in self.locations]

cmaffeo2's avatar
cmaffeo2 committed
504
    def insert_dna(self, at_nt: int, num_nt: int, seq=tuple()):
505
506
507
508
509
510
511
512
513
514
515
516
        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
517
518
        ## TODO: handle sequence

519
520
521
522
523
        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)

524
    def remove_dna(self, first_nt:int, last_nt:int, remove_locations:bool = False):
cmaffeo2's avatar
cmaffeo2 committed
525
        """ Removes nucleotides between first_nt and last_nt, inclusive """
526
527
528
529
530
531
532
533
534
535
536
537
538
        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

539
540
541
542
543
544
545
546
547
        def _transform_contour_positions(contours):
            nt = self.contour_to_nt_pos(contours)
            first = first_nt if first_nt > 0 else -np.inf
            last = last_nt if last_nt < self.num_nt-1 else np.inf
            bad_ids = (nt >= first) * (nt <= last)
            nt[nt>last] = nt[nt>last]-removed_nt
            nt[bad_ids] = np.nan
            return self.nt_pos_to_contour(nt)* self.num_nt/num_nt

548
549
550
        first_nt = np.around(first_nt)
        last_nt = np.around(last_nt)

cmaffeo2's avatar
cmaffeo2 committed
551
552
553
        removed_nt = last_nt-first_nt+1
        num_nt = self.num_nt-removed_nt

554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
        c = np.array([l.address for l in self.locations])
        new_c = _transform_contour_positions(c)

        if np.any(np.isnan(new_c)) and not remove_locations:
            raise Exception("Attempted to remove DNA containing locations from {} between {} and {}".format(self,first_nt,last_nt))

        for l,v in zip(list(self.locations),new_c):
            if np.isnan(v):
                l.delete()
            else:
                l.address = v

        assert( len(self.locations) == np.logical_not(np.isnan(new_c)).sum() )
        tmp_nt = np.array([l.address*num_nt+0.5 for l in self.locations])
        assert(np.all( (tmp_nt < 1) + (tmp_nt > num_nt-1) + np.isclose((np.round(tmp_nt) - tmp_nt), 0) ))

cmaffeo2's avatar
cmaffeo2 committed
570
571
572
573
        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 )
574

575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
        ## Rescale splines
        def _rescale_splines(u, get_coord_function, set_spline_function):
            new_u = _transform_contour_positions(u)
            ids = np.logical_not(np.isnan(new_u))
            new_u = new_u[ids]
            new_p = get_coord_function(u[ids])
            set_spline_function( new_u, new_p )

        _rescale_splines(self.position_spline_params[1],
                         self.contour_to_position,
                         self.set_splines)

        if self.quaternion_spline_params is not None:
            def _contour_to_quaternion(contours):
                os = [self.contour_to_orientation(c) for c in contours]
                qs = [quaternion_from_matrix( o ) for o in os]
                return np.array(qs)
            _rescale_splines(self.quaternion_spline_params[1],
                             self.contour_to_quaternion,
                             self.set_splines)

cmaffeo2's avatar
cmaffeo2 committed
596
        self.num_nt = num_nt
597

598
599
600
601
    def delete(self):
        for c,loc,other in self.get_connections_and_locations():
            c.delete()
        self.parent.segments.remove(self)
602

603
604
605
606
607
608
609
610
611
612
613
614
    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

615
    def translate(self, translation_vector, position_filter=None, contour_filter=None):
616
617
618
619
620
        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)
cmaffeo2's avatar
cmaffeo2 committed
621
        if len(ids) == 0: return
622
623
624

        ## Translate
        r[ids,:] = r[ids,:] + dr[np.newaxis,:]
625
        self.set_splines(u,r)
626
627
628
629
630
631

    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)
cmaffeo2's avatar
cmaffeo2 committed
632
        if len(ids) == 0: return
633
634
635

        if about is None:
            ## TODO: do this more efficiently
636
            r[ids,:] = np.array([rotation_matrix.dot(r[i,:]) for i in ids])
637
638
639
        else:
            dr = np.array(about)
            ## TODO: do this more efficiently
640
            r[ids,:] = np.array([rotation_matrix.dot(r[i,:]-dr) + dr for i in ids])
641

642
        self.set_splines(u,r)
643
644
645
646

        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
647
            orientations = np.array([self.contour_to_orientation(v) for v in u])
648
649
            for i in ids:
                orientations[i,:] = rotation_matrix.dot(orientations[i])
650
            quats = np.array([quaternion_from_matrix(o) for o in orientations])
651
            self.set_orientation_splines(u, quats)
652

653
    def _set_splines_from_ends(self, resolution=4):
654
        self.quaternion_spline_params = None
655
656
657
658
659
660
        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)
661

662
663
664
    def clear_all(self):
        Group.clear_all(self)  # TODO: use super?
        self.beads = []
665
666
667
668
669
        # for c,loc,other in self.get_connections_and_locations():
        #     loc.particle = None
        #     other.particle = None
        for l in self.locations:
            l.particle = None
670

671
    def contour_to_nt_pos(self, contour_pos, round_nt=False):
672
        nt = contour_pos*(self.num_nt) - 0.5
673
        if round_nt:
cmaffeo2's avatar
cmaffeo2 committed
674
            assert( np.isclose(np.around(nt),nt) )
675
676
677
            nt = np.around(nt)
        return nt

678
    def nt_pos_to_contour(self,nt_pos):
679
        return (nt_pos+0.5)/(self.num_nt)
680

681
    def contour_to_position(self,s):
682
        p = interpolate.splev( s, self.position_spline_params[0] )
683
684
685
686
        if len(p) > 1: p = np.array(p).T
        return p

    def contour_to_tangent(self,s):
687
        t = interpolate.splev( s, self.position_spline_params[0], der=1 )
688
689
        t = (t / np.linalg.norm(t,axis=0))
        return t.T
690
691
692
        

    def contour_to_orientation(self,s):
693
        assert( isinstance(s,float) or isinstance(s,int) or len(s) == 1 )   # TODO make vectorized version
694

695
696
697
698
699
        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)
700
            zAxis = np.array((0,0,1))
701
702

            if rotAxisL > 0.001:
703
704
705
                theta = np.arcsin(rotAxisL) * 180/np.pi
                if axis.dot(zAxis) < 0: theta = 180-theta
                orientation0 = rotationAboutAxis( rotAxis/rotAxisL, theta, normalizeAxis=False ).T
706
            else:
707
708
                orientation0 = np.eye(3) if axis.dot(zAxis) > 0 else \
                               rotationAboutAxis( np.array((1,0,0)), 180, normalizeAxis=False )
709
710
711
            if self.start_orientation is not None:
                orientation0 = orientation0.dot(self.start_orientation)

712
713
714
            orientation = rotationAboutAxis( axis, self.twist_per_nt*self.contour_to_nt_pos(s), normalizeAxis=False )
            orientation = orientation.dot(orientation0)
        else:
715
            q = interpolate.splev( s, self.quaternion_spline_params[0] )
716
717
            if len(q) > 1: q = np.array(q).T # TODO: is this needed?
            orientation = quaternion_to_matrix(q)
718

719
        return orientation
720

721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
    def _ntpos_to_seg_and_ntpos(self, nt_pos, is_fwd=True, visited_segs=tuple()):
        """ Cross intrahelical to obtain a tuple of the segment nucleotide position """
        """ TODO: This function could perhaps replace parts of SegmentParticle.get_contour_position """

        if nt_pos >= 0 and nt_pos < self.num_nt:
            return (self,nt_pos,is_fwd)
        else:
            c,A,B = self.get_contour_sorted_connections_and_locations(type_='intrahelical')[0 if nt_pos < 0 else -1]
            ## Special logic for circular segments
            assert(A.container == self)
            if A.container == B.container:
                if (nt_pos < 0 and A.address > B.address) or (nt_pos >= self.num_nt and A.address < B.address):
                    A,B = (B,A)

            other_seg = B.container
            if A.address == (0 if nt_pos < 0 else 1) and other_seg not in visited_segs:
                ## Cross the crossover
                other_pos = (nt_pos-self.num_nt) if nt_pos >= self.num_nt else -nt_pos-1
                if B.address > 0.5:
                    other_pos = (other_seg.num_nt-1) - other_pos
                other_fwd = not is_fwd if A.address == B.address else is_fwd
                return other_seg._ntpos_to_seg_and_ntpos(other_pos, other_fwd,
                                                         list(visited_segs)+[self])

            else:
                ## Could not find a path to nt_pos
                return None
        assert(False)


cmaffeo2's avatar
cmaffeo2 committed
751
    def get_contour_sorted_connections_and_locations(self,type_):
cmaffeo2's avatar
cmaffeo2 committed
752
        sort_fn = lambda c: c[1].address
cmaffeo2's avatar
cmaffeo2 committed
753
        cl = self.get_connections_and_locations(type_)
cmaffeo2's avatar
cmaffeo2 committed
754
        return sorted(cl, key=sort_fn)
755
756
757
    
    def randomize_unset_sequence(self):
        bases = list(seqComplement.keys())
758
        # bases = ['T']        ## FOR DEBUG
759
        if self.sequence is None:
760
            self.sequence = [random.choice(bases) for i in range(self.num_nt)]
761
        else:
762
            assert(len(self.sequence) == self.num_nt) # TODO move
763
764
765
            for i in range(len(self.sequence)):
                if self.sequence[i] is None:
                    self.sequence[i] = random.choice(bases)
766

cmaffeo2's avatar
cmaffeo2 committed
767
768
769
    def _get_num_beads(self, max_basepairs_per_bead, max_nucleotides_per_bead ):
        raise NotImplementedError

770
    def _generate_one_bead(self, contour_position, nts):
771
772
        raise NotImplementedError

cmaffeo2's avatar
cmaffeo2 committed
773
    def _generate_atomic_nucleotide(self, contour_position, is_fwd, seq, scale, strand_segment):
cmaffeo2's avatar
cmaffeo2 committed
774
775
776
777
778
        """ 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)
779
        orientation = self.contour_to_orientation(contour_position)
cmaffeo2's avatar
cmaffeo2 committed
780

781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
        """ 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
799
        key = seq
800
801
        nt_dict = canonicalNtFwd if is_fwd else canonicalNtRev

802
        atoms = nt_dict[ key ].generate() # TODO: clone?
cmaffeo2's avatar
cmaffeo2 committed
803
        atoms.orientation = orientation.dot(atoms.orientation)
804
805
806
807
        if isinstance(self, SingleStrandedSegment):
            if scale is not None and scale != 1:
                for a in atoms:
                    a.position = scale*a.position
808
            atoms.position = pos - atoms.atoms_by_name["C1'"].collapsedPosition()
809
810
811
812
813
814
815
816
817
        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
818
                    else:
819
                        a.fixed = 1
820
            atoms.position = pos
cmaffeo2's avatar
cmaffeo2 committed
821
822
823
824
825
826
        
        atoms.contour_position = contour_position
        strand_segment.add(atoms)

        for callback in self._generate_nucleotide_callbacks:
            callback(atoms)
cmaffeo2's avatar
cmaffeo2 committed
827
828

        return atoms
829

830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
    def _generate_oxdna_nucleotide(self, contour_position, is_fwd, seq):
        bp_center = self.contour_to_position(contour_position)
        orientation = self.contour_to_orientation(contour_position)

        DefaultOrientation = rotationAboutAxis([0,0,1], 90)
        if is_fwd: 
            DefaultOrientation = rotationAboutAxis([1,0,0], 180).dot(DefaultOrientation)

        o = orientation.dot(DefaultOrientation)

        if isinstance(self, SingleStrandedSegment):
            pos = bp_center
        else:
            pos = bp_center - 5*o.dot(np.array((1,0,0)))

        nt = PointParticle("oxdna_nt", position= pos,
                             orientation = o)

        nt.contour_position = contour_position
        return nt


852
853
    def add_location(self, nt, type_, on_fwd_strand=True):
        ## Create location if needed, add to segment
854
        c = self.nt_pos_to_contour(nt)
855
856
857
858
859
860
        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?
861
862

    def add_nick(self, nt, on_fwd_strand=True):
863
864
865
866
867
868
869
        if on_fwd_strand:
            self.add_3prime(nt,on_fwd_strand)
            self.add_5prime(nt+1,on_fwd_strand)
        else:
            self.add_5prime(nt,on_fwd_strand)
            self.add_3prime(nt+1,on_fwd_strand)

870

871
    def add_5prime(self, nt, on_fwd_strand=True):
872
873
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
874
        self.add_location(nt,"5prime",on_fwd_strand)
875
876

    def add_3prime(self, nt, on_fwd_strand=True):
877
878
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
879
        self.add_location(nt,"3prime",on_fwd_strand)
880

881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
    ## Real work
    def _connect_ends(self, end1, end2, type_, force_connection=False):
        debug = False
        ## TODO remove self?
        ## validate the input
        for end in (end1, end2):
            assert( isinstance(end, Location) )
            assert( end.type_ in ("end3","end5") )
        assert( end1.type_ != end2.type_ )

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

        ## Create and add connection
        if end2.type_ == "end5":
            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 )

905
    def get_3prime_locations(self):
cmaffeo2's avatar
cmaffeo2 committed
906
        return sorted(self.get_locations("3prime"),key=lambda x: x.address)
907
    
cmaffeo2's avatar
cmaffeo2 committed
908
    def get_5prime_locations(self):
909
        ## TODO? ensure that data is consistent before _build_model calls
cmaffeo2's avatar
cmaffeo2 committed
910
        return sorted(self.get_locations("5prime"),key=lambda x: x.address)
cmaffeo2's avatar
cmaffeo2 committed
911

912
    def iterate_connections_and_locations(self, reverse=False):
cmaffeo2's avatar
cmaffeo2 committed
913
914
        ## connections to other segments
        cl = self.get_contour_sorted_connections_and_locations()
915
        if reverse:
cmaffeo2's avatar
cmaffeo2 committed
916
            cl = cl[::-1]
917
918
919
            
        for c in cl:
            yield c
cmaffeo2's avatar
cmaffeo2 committed
920

921
922
923
924
925
926
927
928
929
930
931
    ## 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):
932
933
                if value >= l and value <= h:
                    raise Exception("Strand piece already exists! DNA may be circular.")
934
935
936
937
938
939

        ## 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)

940
    ## TODO rename
941
    def get_strand_segment(self, nt_pos, is_fwd, move_at_least=0.5):
942
        """ Walks through locations, checking for crossovers """
943
944
945
946
        # if self.name in ("6-1","1-1"):
        #     import pdb
        #     pdb.set_trace()
        move_at_least = 0
947
948

        ## Iterate through locations
cmaffeo2's avatar
cmaffeo2 committed
949
        # locations = sorted(self.locations, key=lambda l:(l.address,not l.on_fwd_strand), reverse=(not is_fwd))
950
951
952
953
954
955
        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))
956
957
        # print(locations)

958
        for l in locations:
cmaffeo2's avatar
cmaffeo2 committed
959
960
961
962
            # TODOTODO probably okay
            if l.address == 0:
                pos = 0.0
            elif l.address == 1:
963
                pos = self.num_nt-1
cmaffeo2's avatar
cmaffeo2 committed
964
965
            else:
                pos = self.contour_to_nt_pos(l.address, round_nt=True)
966
967
968

            ## DEBUG

cmaffeo2's avatar
cmaffeo2 committed
969

970
            ## Skip locations encountered before our strand
971
972
973
974
975
976
977
978
            # 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
979
980

            ## Stop if we found the 3prime end
981
            if l.on_fwd_strand == is_fwd and l.type_ == "3prime" and l.connection is None:
982
983
                if _DEBUG_TRACE:
                    print("  found end at",l)
984
                return pos, None, None, None, None
985
986
987
988
989
990
991
992

            ## 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:
993
994
995
                if _DEBUG_TRACE:
                    print("  passing through",l)
                    print("from {}, connection {} to {}".format(nt_pos,l,B))
996
                Bpos = B.get_nt_pos()
997
                return pos, B.container, Bpos, B.on_fwd_strand, 0.5
998
999
1000
                
            ## Stop at other strand crossovers so basepairs line up
            elif c.type_ == "crossover":
For faster browsing, not all history is shown. View entire blame