segmentmodel.py 149 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 .arbdmodel import PointParticle, ParticleType, Group, ArbdModel
from .arbdmodel.coords import rotationAboutAxis, quaternion_from_matrix, quaternion_to_matrix
from .arbdmodel.nonbonded import *
8
from copy import copy, deepcopy
9
from .model.nbPot import nbDnaScheme
10

11
12
import re

cmaffeo2's avatar
cmaffeo2 committed
13
14
from scipy.special import erf
import scipy.optimize as opt
15
from scipy import interpolate
cmaffeo2's avatar
cmaffeo2 committed
16

17
18
from .model.CanonicalNucleotideAtoms import canonicalNtFwd, canonicalNtRev, seqComplement
from .model.CanonicalNucleotideAtoms import enmTemplateHC, enmTemplateSQ, enmCorrectionsHC
cmaffeo2's avatar
cmaffeo2 committed
19

cmaffeo2's avatar
cmaffeo2 committed
20
from .model.spring_from_lp import k_angle as angle_spring_from_lp
21
from .model.spring_from_lp import k_twist as twist_spring_from_lp
cmaffeo2's avatar
cmaffeo2 committed
22

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
"""
cmaffeo2's avatar
cmaffeo2 committed
50
51
class CircularDnaError(Exception):
    pass
52

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

56
57
class Location():
    """ Site for connection within an object """
58
    def __init__(self, container, address, type_, on_fwd_strand = True):
59
        ## TODO: remove cyclic references(?)
60
        self.container = container
cmaffeo2's avatar
cmaffeo2 committed
61
        self.address = address  # represents position along contour length in segment
cmaffeo2's avatar
cmaffeo2 committed
62
        # assert( type_ in ("end3","end5") ) # TODO remove or make conditional
63
        self.on_fwd_strand = on_fwd_strand
64
65
        self.type_ = type_
        self.particle = None
66
        self.connection = None
67
        self.is_3prime_side_of_connection = None
68

69
70
        self.prev_in_strand = None
        self.next_in_strand = 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
102
    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)
        
103
104
105
106
107
108
109
110
111
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_
        
112
113
114
115
116
117
118
    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
119

120
121
    def delete(self):
        self.A.container.connections.remove(self)
122
123
        if self.B.container is not self.A.container:
            self.B.container.connections.remove(self)
124
125
126
        self.A.connection = None
        self.B.connection = None

cmaffeo2's avatar
cmaffeo2 committed
127
128
129
130
    def __repr__(self):
        return "<Connection {}--{}--{}]>".format( self.A, self.type_, self.B )
        

131
        
132
133
134
# class ConnectableElement(Transformable):
class ConnectableElement():
    """ Abstract base class """
cmaffeo2's avatar
cmaffeo2 committed
135
136
137
138
    def __init__(self, connection_locations=None, connections=None):
        if connection_locations is None: connection_locations = []
        if connections is None: connections = []

139
140
        ## TODO decide on names
        self.locations = self.connection_locations = connection_locations
141
142
        self.connections = connections

cmaffeo2's avatar
cmaffeo2 committed
143
    def get_locations(self, type_=None, exclude=()):
144
145
146
147
148
149
150
151
        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] ) )
152
153
154
155
        return locs

    def get_location_at(self, address, on_fwd_strand=True, new_type="crossover"):
        loc = None
156
        if (self.num_nt == 1):
157
158
159
160
161
162
163
            # 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
164
            # assert( loc is not None )
165
166
167
168
169
170
171
172
        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
173

cmaffeo2's avatar
cmaffeo2 committed
174
    def get_connections_and_locations(self, connection_type=None, exclude=()):
175
176
        """ Returns a list with each entry of the form:
            connection, location_in_self, location_in_other """
177
        type_ = connection_type
178
179
        ret = []
        for c in self.connections:
180
            if (type_ is None or c.type_ == type_) and c.type_ not in exclude:
181
                if   c.A.container is self:
182
                    ret.append( [c, c.A, c.B] )
183
                elif c.B.container is self:
184
185
                    ret.append( [c, c.B, c.A] )
                else:
186
187
                    import pdb
                    pdb.set_trace()
188
189
190
                    raise Exception("Object contains connection that fails to refer to object")
        return ret

191
    def _connect(self, other, connection, in_3prime_direction=None):
192
193
        ## TODO fix circular references        
        A,B = [connection.A, connection.B]
194
195
196
197
        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
            
198
        A.connection = B.connection = connection
199
        self.connections.append(connection)
200
201
        if other is not self:
            other.connections.append(connection)
cmaffeo2's avatar
cmaffeo2 committed
202
203
        else:
            raise NotImplementedError("Segments cannot yet be connected to themselves; if you are attempting to make a circular object, try breaking the object into multiple segments")
204
205
206
207
208
209
        l = A.container.locations
        if A not in l: l.append(A)
        l = B.container.locations
        if B not in l: l.append(B)
        

210
211
    # def _find_connections(self, loc):
    #     return [c for c in self.connections if c.A == loc or c.B == loc]
212
213

class SegmentParticle(PointParticle):
cmaffeo2's avatar
cmaffeo2 committed
214
    def __init__(self, type_, position, name="A", **kwargs):
215
        self.name = name
216
        self.contour_position = None
cmaffeo2's avatar
cmaffeo2 committed
217
        PointParticle.__init__(self, type_, position, name=name, **kwargs)
218
219
        self.intrahelical_neighbors = []
        self.other_neighbors = []
cmaffeo2's avatar
cmaffeo2 committed
220
        self.locations = []
221

222
    def get_intrahelical_above(self, all_types=True):
223
        """ Returns bead directly above self """
224
        # assert( len(self.intrahelical_neighbors) <= 2 )
225
        for b in self.intrahelical_neighbors:
226
            if b.get_contour_position(self.parent, self.contour_position) > self.contour_position:
227
228
                if all_types or isinstance(b,type(self)):
                    return b
229

230
    def get_intrahelical_below(self, all_types=True):
231
        """ Returns bead directly below 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

cmaffeo2's avatar
cmaffeo2 committed
238
    def _neighbor_should_be_added(self,b):
239
240
241
        if type(self.parent) != type(b.parent):
            return True

cmaffeo2's avatar
cmaffeo2 committed
242
        c1 = self.contour_position
243
        c2 = b.get_contour_position(self.parent,c1)
cmaffeo2's avatar
cmaffeo2 committed
244
245
        if c2 < c1:
            b0 = self.get_intrahelical_below()
246
        else:
cmaffeo2's avatar
cmaffeo2 committed
247
248
249
            b0 = self.get_intrahelical_above()

        if b0 is not None:            
250
            c0 = b0.get_contour_position(self.parent,c1)
cmaffeo2's avatar
cmaffeo2 committed
251
252
253
254
255
256
257
258
259
260
261
262
263
            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:
264
265
            # assert(len(b.intrahelical_neighbors) <= 1)
            # assert(len(self.intrahelical_neighbors) <= 1)
cmaffeo2's avatar
cmaffeo2 committed
266
267
            self.intrahelical_neighbors.append(b)
            b.intrahelical_neighbors.append(self)
268

269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
    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
295
296
297
298
        """ Returns the "address" of the nucleotide relative to seg in
        nucleotides, taking the shortest (intrahelical) contour length route to seg
        """
        if seg == self.parent:
299
            pos = self.contour_position
cmaffeo2's avatar
cmaffeo2 committed
300
        else:
301
302
            pos = self.get_contour_position(seg,near_address)
        return seg.contour_to_nt_pos(pos)
cmaffeo2's avatar
cmaffeo2 committed
303

304
    def get_contour_position(self,seg, address = None):
cmaffeo2's avatar
cmaffeo2 committed
305
306
307
308
        """ 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
        """

309
310
311
        if seg == self.parent:
            return self.contour_position
        else:
cmaffeo2's avatar
cmaffeo2 committed
312
313
314
315
316
            cutoff = 30*3
            target_seg = seg

            ## depth-first search
            ## TODO cache distances to nearby locations?
cmaffeo2's avatar
cmaffeo2 committed
317
            def descend_search_tree(seg, contour_in_seg, distance=0, visited_segs=None):
cmaffeo2's avatar
cmaffeo2 committed
318
                nonlocal cutoff
cmaffeo2's avatar
cmaffeo2 committed
319
                if visited_segs is None: visited_segs = []
cmaffeo2's avatar
cmaffeo2 committed
320
321
322
323

                if seg == target_seg:
                    # pdb.set_trace()
                    ## Found a segment in our target
324
325
                    sign = 1 if contour_in_seg == 1 else -1
                    if sign == -1: assert( contour_in_seg == 0 )
cmaffeo2's avatar
cmaffeo2 committed
326
327
                    if distance < cutoff: # TODO: check if this does anything
                        cutoff = distance
328
                    return [[distance, contour_in_seg+sign*seg.nt_pos_to_contour(distance)]], [(seg, contour_in_seg, distance)]
cmaffeo2's avatar
cmaffeo2 committed
329
                if distance > cutoff:
330
                    return None,None
cmaffeo2's avatar
cmaffeo2 committed
331
332
                    
                ret_list = []
333
                hist_list = []
cmaffeo2's avatar
cmaffeo2 committed
334
                ## Find intrahelical locations in seg that we might pass through
335
336
337
338
339
                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
340
                    if B.container in visited_segs: continue
341
342
343
                    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
344
345
346
                                                   distance+dx, visited_segs + [seg] )
                    if results is not None:
                        ret_list.extend( results )
347
348
                        hist_list.extend( history )
                return ret_list,hist_list
cmaffeo2's avatar
cmaffeo2 committed
349

350
            results,history = descend_search_tree(self.parent, self.contour_position)
cmaffeo2's avatar
cmaffeo2 committed
351
352
            if results is None or len(results) == 0:
                raise Exception("Could not find location in segment") # TODO better error
353
354
355
356
            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
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
            # 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
372
373
374
    def __repr__(self):
        return "<SegmentParticle {} on {}[{:.2f}]>".format( self.name, self.parent, self.contour_position)

375
376

## TODO break this class into smaller, better encapsulated pieces
377
378
379
380
381
382
383
384
385
386
387
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
388
389
390
391
392
    orientation_particle = ParticleType("O",
                                        diffusivity = 100,
                                        mass = 300,
                                        radius = 1,
                                    )
393

cmaffeo2's avatar
cmaffeo2 committed
394
    # orientation_bond = HarmonicBond(10,2)
395
    orientation_bond = HarmonicBond(30,1.5, rRange = (0,500) )
396
397
398
399
400
401
402

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

403
    def __init__(self, name, num_nt, 
cmaffeo2's avatar
cmaffeo2 committed
404
                 start_position = None,
405
                 end_position = None, 
cmaffeo2's avatar
cmaffeo2 committed
406
407
                 segment_model = None,
                 **kwargs):
408

cmaffeo2's avatar
cmaffeo2 committed
409
410
        if start_position is None: start_position = np.array((0,0,0))

cmaffeo2's avatar
cmaffeo2 committed
411
        Group.__init__(self, name, children=[], **kwargs)
412
        ConnectableElement.__init__(self, connection_locations=[], connections=[])
413

cmaffeo2's avatar
cmaffeo2 committed
414
415
416
        if 'segname' not in kwargs:
            self.segname = name
        # self.resname = name
cmaffeo2's avatar
cmaffeo2 committed
417
418
419
420
421
        self.start_orientation = None
        self.twist_per_nt = 0

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

422
423
424
        self._bead_model_generation = 0    # TODO: remove?
        self.segment_model = segment_model # TODO: remove?

425
426
427
428
        self.strand_pieces = dict()
        for d in ('fwd','rev'):
            self.strand_pieces[d] = []

429
        self.num_nt = int(num_nt)
430
        if end_position is None:
431
            end_position = np.array((0,0,self.distance_per_nt*num_nt)) + start_position
432
433
434
        self.start_position = start_position
        self.end_position = end_position

cmaffeo2's avatar
cmaffeo2 committed
435
436
437
438
        ## Used to assign cadnano names to beads
        self._generate_bead_callbacks = []
        self._generate_nucleotide_callbacks = []

439
        ## Set up interpolation for positions
440
441
442
443
        self._set_splines_from_ends()

        self.sequence = None

444
    def __repr__(self):
445
        return "<{} {}[{:d}]>".format( str(type(self)).split('.')[-1], self.name, self.num_nt )
446

cmaffeo2's avatar
cmaffeo2 committed
447
    def set_splines(self, contours, coords):
448
        tck, u = interpolate.splprep( coords.T, u=contours, s=0, k=1)
449
        self.position_spline_params = (tck,u)
450

451
452
    def set_orientation_splines(self, contours, quaternions):
        tck, u = interpolate.splprep( quaternions.T, u=contours, s=0, k=1)
453
454
455
456
457
458
        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)

459
460
461
462
    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):
463
            positions[i] = self.contour_to_position(c)
464
465
466
467
468
            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_

469
470
471
    def _get_location_positions(self):
        return [self.contour_to_nt_pos(l.address) for l in self.locations]

cmaffeo2's avatar
cmaffeo2 committed
472
    def insert_dna(self, at_nt: int, num_nt: int, seq=tuple()):
473
474
475
476
477
478
479
480
481
482
483
484
        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
485
486
        ## TODO: handle sequence

487
488
489
490
491
492
        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
493
        """ Removes nucleotides between first_nt and last_nt, inclusive """
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
        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
516
517
518
519
520
521
522
523
        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 )
524

cmaffeo2's avatar
cmaffeo2 committed
525
        self.num_nt = num_nt
526
527
528
529

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

530
531
532
533
534
535
536
537
538
539
540
541
    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

542
    def translate(self, translation_vector, position_filter=None, contour_filter=None):
543
544
545
546
547
        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
548
        if len(ids) == 0: return
549
550
551

        ## Translate
        r[ids,:] = r[ids,:] + dr[np.newaxis,:]
552
        self.set_splines(u,r)
553
554
555
556
557
558

    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
559
        if len(ids) == 0: return
560
561
562

        if about is None:
            ## TODO: do this more efficiently
563
            r[ids,:] = np.array([rotation_matrix.dot(r[i,:]) for i in ids])
564
565
566
        else:
            dr = np.array(about)
            ## TODO: do this more efficiently
567
            r[ids,:] = np.array([rotation_matrix.dot(r[i,:]-dr) + dr for i in ids])
568

569
        self.set_splines(u,r)
570
571
572
573

        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
574
            orientations = np.array([self.contour_to_orientation(v) for v in u])
575
576
577
            for i in ids:
                orientations[i,:] = rotation_matrix.dot(orientations[i])
            quats = [quaternion_from_matrix(o) for o in orientations]
578
            self.set_orientation_splines(u, quats)
579

580
    def _set_splines_from_ends(self, resolution=4):
581
        self.quaternion_spline_params = None
582
583
584
585
586
587
        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)
588

589
590
591
    def clear_all(self):
        Group.clear_all(self)  # TODO: use super?
        self.beads = []
592
593
594
595
596
        # for c,loc,other in self.get_connections_and_locations():
        #     loc.particle = None
        #     other.particle = None
        for l in self.locations:
            l.particle = None
597

598
    def contour_to_nt_pos(self, contour_pos, round_nt=False):
599
        nt = contour_pos*(self.num_nt) - 0.5
600
        if round_nt:
cmaffeo2's avatar
cmaffeo2 committed
601
            assert( np.isclose(np.around(nt),nt) )
602
603
604
            nt = np.around(nt)
        return nt

605
    def nt_pos_to_contour(self,nt_pos):
606
        return (nt_pos+0.5)/(self.num_nt)
607

608
    def contour_to_position(self,s):
609
        p = interpolate.splev( s, self.position_spline_params[0] )
610
611
612
613
        if len(p) > 1: p = np.array(p).T
        return p

    def contour_to_tangent(self,s):
614
        t = interpolate.splev( s, self.position_spline_params[0], der=1 )
615
616
        t = (t / np.linalg.norm(t,axis=0))
        return t.T
617
618
619
        

    def contour_to_orientation(self,s):
620
        assert( isinstance(s,float) or isinstance(s,int) or len(s) == 1 )   # TODO make vectorized version
621

622
623
624
625
626
        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)
627
            zAxis = np.array((0,0,1))
628
629

            if rotAxisL > 0.001:
630
631
632
                theta = np.arcsin(rotAxisL) * 180/np.pi
                if axis.dot(zAxis) < 0: theta = 180-theta
                orientation0 = rotationAboutAxis( rotAxis/rotAxisL, theta, normalizeAxis=False ).T
633
            else:
634
635
                orientation0 = np.eye(3) if axis.dot(zAxis) > 0 else \
                               rotationAboutAxis( np.array((1,0,0)), 180, normalizeAxis=False )
636
637
638
            if self.start_orientation is not None:
                orientation0 = orientation0.dot(self.start_orientation)

639
640
641
            orientation = rotationAboutAxis( axis, self.twist_per_nt*self.contour_to_nt_pos(s), normalizeAxis=False )
            orientation = orientation.dot(orientation0)
        else:
642
            q = interpolate.splev( s, self.quaternion_spline_params[0] )
643
644
            if len(q) > 1: q = np.array(q).T # TODO: is this needed?
            orientation = quaternion_to_matrix(q)
645

646
        return orientation
647

cmaffeo2's avatar
cmaffeo2 committed
648
    def get_contour_sorted_connections_and_locations(self,type_):
cmaffeo2's avatar
cmaffeo2 committed
649
        sort_fn = lambda c: c[1].address
cmaffeo2's avatar
cmaffeo2 committed
650
        cl = self.get_connections_and_locations(type_)
cmaffeo2's avatar
cmaffeo2 committed
651
        return sorted(cl, key=sort_fn)
652
653
654
    
    def randomize_unset_sequence(self):
        bases = list(seqComplement.keys())
655
        # bases = ['T']        ## FOR DEBUG
656
        if self.sequence is None:
657
            self.sequence = [random.choice(bases) for i in range(self.num_nt)]
658
        else:
659
            assert(len(self.sequence) == self.num_nt) # TODO move
660
661
662
            for i in range(len(self.sequence)):
                if self.sequence[i] is None:
                    self.sequence[i] = random.choice(bases)
663

cmaffeo2's avatar
cmaffeo2 committed
664
665
666
    def _get_num_beads(self, max_basepairs_per_bead, max_nucleotides_per_bead ):
        raise NotImplementedError

667
    def _generate_one_bead(self, contour_position, nts):
668
669
        raise NotImplementedError

cmaffeo2's avatar
cmaffeo2 committed
670
    def _generate_atomic_nucleotide(self, contour_position, is_fwd, seq, scale, strand_segment):
cmaffeo2's avatar
cmaffeo2 committed
671
672
673
674
675
        """ 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)
676
        orientation = self.contour_to_orientation(contour_position)
cmaffeo2's avatar
cmaffeo2 committed
677

678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
        """ 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
696
        key = seq
697
698
        nt_dict = canonicalNtFwd if is_fwd else canonicalNtRev

699
        atoms = nt_dict[ key ].generate() # TODO: clone?
cmaffeo2's avatar
cmaffeo2 committed
700
        atoms.orientation = orientation.dot(atoms.orientation)
701
702
703
704
        if isinstance(self, SingleStrandedSegment):
            if scale is not None and scale != 1:
                for a in atoms:
                    a.position = scale*a.position
705
            atoms.position = pos - atoms.atoms_by_name["C1'"].collapsedPosition()
706
707
708
709
710
711
712
713
714
        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
715
                    else:
716
                        a.fixed = 1
717
            atoms.position = pos
cmaffeo2's avatar
cmaffeo2 committed
718
719
720
721
722
723
        
        atoms.contour_position = contour_position
        strand_segment.add(atoms)

        for callback in self._generate_nucleotide_callbacks:
            callback(atoms)
cmaffeo2's avatar
cmaffeo2 committed
724
725

        return atoms
726

727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
    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


749
750
    def add_location(self, nt, type_, on_fwd_strand=True):
        ## Create location if needed, add to segment
751
        c = self.nt_pos_to_contour(nt)
752
753
754
755
756
757
        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?
758
759

    def add_nick(self, nt, on_fwd_strand=True):
760
761
762
763
764
765
766
        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)

767

768
    def add_5prime(self, nt, on_fwd_strand=True):
769
770
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
771
        self.add_location(nt,"5prime",on_fwd_strand)
772
773

    def add_3prime(self, nt, on_fwd_strand=True):
774
775
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
776
        self.add_location(nt,"3prime",on_fwd_strand)
777

778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
    ## 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 )

802
    def get_3prime_locations(self):
cmaffeo2's avatar
cmaffeo2 committed
803
        return sorted(self.get_locations("3prime"),key=lambda x: x.address)
804
    
cmaffeo2's avatar
cmaffeo2 committed
805
    def get_5prime_locations(self):
806
        ## TODO? ensure that data is consistent before _build_model calls
cmaffeo2's avatar
cmaffeo2 committed
807
        return sorted(self.get_locations("5prime"),key=lambda x: x.address)
cmaffeo2's avatar
cmaffeo2 committed
808

809
    def iterate_connections_and_locations(self, reverse=False):
cmaffeo2's avatar
cmaffeo2 committed
810
811
        ## connections to other segments
        cl = self.get_contour_sorted_connections_and_locations()
812
        if reverse:
cmaffeo2's avatar
cmaffeo2 committed
813
            cl = cl[::-1]
814
815
816
            
        for c in cl:
            yield c
cmaffeo2's avatar
cmaffeo2 committed
817

818
819
820
821
822
823
824
825
826
827
828
    ## 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):
829
830
                if value >= l and value <= h:
                    raise Exception("Strand piece already exists! DNA may be circular.")
831
832
833
834
835
836

        ## 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)

837
    ## TODO rename
838
    def get_strand_segment(self, nt_pos, is_fwd, move_at_least=0.5):
839
        """ Walks through locations, checking for crossovers """
840
841
842
843
        # if self.name in ("6-1","1-1"):
        #     import pdb
        #     pdb.set_trace()
        move_at_least = 0
844
845

        ## Iterate through locations
cmaffeo2's avatar
cmaffeo2 committed
846
        # locations = sorted(self.locations, key=lambda l:(l.address,not l.on_fwd_strand), reverse=(not is_fwd))
847
848
849
850
851
852
        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))
853
854
        # print(locations)

855
        for l in locations:
cmaffeo2's avatar
cmaffeo2 committed
856
857
858
859
            # TODOTODO probably okay
            if l.address == 0:
                pos = 0.0
            elif l.address == 1:
860
                pos = self.num_nt-1
cmaffeo2's avatar
cmaffeo2 committed
861
862
            else:
                pos = self.contour_to_nt_pos(l.address, round_nt=True)
863
864
865

            ## DEBUG

cmaffeo2's avatar
cmaffeo2 committed
866

867
            ## Skip locations encountered before our strand
868
869
870
871
872
873
874
875
            # 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
876
877

            ## Stop if we found the 3prime end
878
            if l.on_fwd_strand == is_fwd and l.type_ == "3prime" and l.connection is None:
879
                # print("  found end at",l)
880
                return pos, None, None, None, None
881
882
883
884
885
886
887
888

            ## 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:
889
890
                # print("  passing through",l)
                # print("from {}, connection {} to {}".format(nt_pos,l,B))
891
                Bpos = B.get_nt_pos()
892
                return pos, B.container, Bpos, B.on_fwd_strand, 0.5
893
894
895
                
            ## Stop at other strand crossovers so basepairs line up
            elif c.type_ == "crossover":
896
                if nt_pos == pos: continue
897
                # print("  pausing at",l)
898
                return pos, l.container, pos+(2*is_fwd-1), is_fwd, 0
899
900
901
902
903
904

        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

905
906
907
    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
908
        # TODO: include beads in connections?
909
910
911
        i = np.argmin((cs - contour_position)**2)

        return self.beads[i]
912

913
914
915
916
917
918
919
920
921
    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 ))

922
923
    def get_all_consecutive_beads(self, number):
        assert(number >= 1)
cmaffeo2's avatar
cmaffeo2 committed
924
        ## Assume that consecutive beads in self.beads are bonded
925
        ret = []
cmaffeo2's avatar
cmaffeo2 committed
926
927
        for i in range(len(self.beads)-number+1):
            tmp = [self.beads[i+j] for j in range(0,number)]
928
            ret.append( tmp )
929
        return ret   
930

931
    def _add_bead(self,b):
932
        
933
934
935
        # assert(b.parent is None)
        if b.parent is not None:
            b.parent.children.remove(b)
936
        self.add(b)
937
938
939
940
941
942
        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)
943
            self.add(o)
944
945
946
947
948
949
950
951
            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 = []
952
953

        if True:
954
955
            ## TODO: remove this if duplicates are never found 
            # print("Searching for duplicate particles...")
956
            ## Remove duplicates, preserving order
957
958
959
960
            tmp = []
            for c in new_children:
                if c not in tmp:
                    tmp.append(c)
961
                else:
962
                    print("  DUPLICATE PARTICLE FOUND!")
963
964
            new_children = tmp

965
966
967
968
969
        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)
970
971
972
973
974
            
        # 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)
975
976
        assert(len(old_children) == len(self.children))
        assert(len(old_beads) == len(self.beads))
977

978

cmaffeo2's avatar
cmaffeo2 committed
979
    def _generate_beads(self, bead_model, max_basepairs_per_bead, max_nucleotides_per_bead):
980

981
        """ Generate beads (positions, types, etc) and bonds, angles, dihedrals, exclusions """
cmaffeo2's avatar
cmaffeo2 committed
982
        ## TODO: decide whether to remove bead_model argument
983
        ##       (currently unused)
cmaffeo2's avatar
cmaffeo2 committed
984

985
        ## First find points between-which beads must be generated
986
987
988
        # 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]