segmentmodel.py 164 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
from . import get_resource_path

14
15
import re

cmaffeo2's avatar
cmaffeo2 committed
16
17
from scipy.special import erf
import scipy.optimize as opt
18
from scipy import interpolate
cmaffeo2's avatar
cmaffeo2 committed
19

20
21
from .model.CanonicalNucleotideAtoms import canonicalNtFwd, canonicalNtRev, seqComplement
from .model.CanonicalNucleotideAtoms import enmTemplateHC, enmTemplateSQ, enmCorrectionsHC
cmaffeo2's avatar
cmaffeo2 committed
22

cmaffeo2's avatar
cmaffeo2 committed
23
from .model.spring_from_lp import k_angle as angle_spring_from_lp
24
from .model.spring_from_lp import k_twist as twist_spring_from_lp
cmaffeo2's avatar
cmaffeo2 committed
25

26
27
import csv

28
# import pdb
29
"""
cmaffeo2's avatar
cmaffeo2 committed
30
TODO:
cmaffeo2's avatar
cmaffeo2 committed
31
 + fix handling of crossovers for atomic representation
cmaffeo2's avatar
cmaffeo2 committed
32
 + map to atomic representation
33
    + add nicks
cmaffeo2's avatar
cmaffeo2 committed
34
    + transform ssDNA nucleotides 
cmaffeo2's avatar
cmaffeo2 committed
35
    - shrink ssDNA
cmaffeo2's avatar
cmaffeo2 committed
36
    + shrink dsDNA backbone
37
    + make orientation continuous
cmaffeo2's avatar
cmaffeo2 committed
38
    + sequence
39
    + handle circular dna
40
 + ensure crossover bead potentials aren't applied twice 
41
 + remove performance bottlenecks
42
 - test for large systems
cmaffeo2's avatar
cmaffeo2 committed
43
 + assign sequence
44
 + ENM
45
46
 - rework Location class 
 - remove recursive calls
47
 - document
48
 - develop unit test suite
49
50
 - 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
51
"""
52
53
54

_DEBUG_TRACE = False

cmaffeo2's avatar
cmaffeo2 committed
55
56
class CircularDnaError(Exception):
    pass
57

cmaffeo2's avatar
cmaffeo2 committed
58
59
60
class ParticleNotConnectedError(Exception):
    pass

61
62
class Location():
    """ Site for connection within an object """
63
    def __init__(self, container, address, type_, on_fwd_strand = True):
64
        ## TODO: remove cyclic references(?)
65
        self.container = container
cmaffeo2's avatar
cmaffeo2 committed
66
        self.address = address  # represents position along contour length in segment
cmaffeo2's avatar
cmaffeo2 committed
67
        # assert( type_ in ("end3","end5") ) # TODO remove or make conditional
68
        self.on_fwd_strand = on_fwd_strand
69
70
        self.type_ = type_
        self.particle = None
71
        self.connection = None
72
        self.is_3prime_side_of_connection = None
73

74
        self.combine = None     # some locations might be combined in bead model 
75
76
77
78
79
80
81

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

82
    def set_connection(self, connection, is_3prime_side_of_connection):
83
        self.connection = connection # TODO weakref? 
84
        self.is_3prime_side_of_connection = is_3prime_side_of_connection
85

86
87
88
89
90
91
92
    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:
93
                pos = self.container.num_nt-1
94
95
96
97
            else:
                raise
        return pos

98
99
100
101
102
103
    def delete(self):
        if self.connection is not None:
            self.connection.delete()
        if self.container is not None:
            self.container.locations.remove(self)

104
105
106
107
108
    def __repr__(self):
        if self.on_fwd_strand:
            on_fwd = "on_fwd_strand"
        else:
            on_fwd = "on_rev_strand"
109
110
111
112
        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)
113
        
114
115
116
117
118
119
120
121
122
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_
        
123
124
125
126
127
128
129
    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
130

131
132
    def delete(self):
        self.A.container.connections.remove(self)
133
134
        if self.B.container is not self.A.container:
            self.B.container.connections.remove(self)
135
136
137
        self.A.connection = None
        self.B.connection = None

cmaffeo2's avatar
cmaffeo2 committed
138
139
140
141
    def __repr__(self):
        return "<Connection {}--{}--{}]>".format( self.A, self.type_, self.B )
        

142
        
143
144
145
# class ConnectableElement(Transformable):
class ConnectableElement():
    """ Abstract base class """
cmaffeo2's avatar
cmaffeo2 committed
146
147
148
149
    def __init__(self, connection_locations=None, connections=None):
        if connection_locations is None: connection_locations = []
        if connections is None: connections = []

150
151
        ## TODO decide on names
        self.locations = self.connection_locations = connection_locations
152
153
        self.connections = connections

cmaffeo2's avatar
cmaffeo2 committed
154
    def get_locations(self, type_=None, exclude=()):
155
156
157
158
159
160
161
162
        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] ) )
163
164
        return locs

165
    def get_location_at(self, address, on_fwd_strand=True, new_type=None):
166
        loc = None
167
        if (self.num_nt == 1):
168
169
170
171
172
173
174
            # 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
175
            # assert( loc is not None )
176
177
178
179
180
        else:
            for l in self.locations:
                if l.address == address and l.on_fwd_strand == on_fwd_strand:
                    assert(loc is None)
                    loc = l
181
        if loc is None and new_type is not None:
182
183
            loc = Location( self, address=address, type_=new_type, on_fwd_strand=on_fwd_strand )
        return loc
184

cmaffeo2's avatar
cmaffeo2 committed
185
    def get_connections_and_locations(self, connection_type=None, exclude=()):
186
187
        """ Returns a list with each entry of the form:
            connection, location_in_self, location_in_other """
188
        type_ = connection_type
189
190
        ret = []
        for c in self.connections:
191
            if (type_ is None or c.type_ == type_) and c.type_ not in exclude:
192
                if   c.A.container is self:
193
                    ret.append( [c, c.A, c.B] )
194
                elif c.B.container is self:
195
196
                    ret.append( [c, c.B, c.A] )
                else:
197
198
                    import pdb
                    pdb.set_trace()
199
200
201
                    raise Exception("Object contains connection that fails to refer to object")
        return ret

202
    def _connect(self, other, connection, in_3prime_direction=None):
203
204
        ## TODO fix circular references        
        A,B = [connection.A, connection.B]
205
206
207
208
        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
            
209
        A.connection = B.connection = connection
210
        self.connections.append(connection)
211
212
        if other is not self:
            other.connections.append(connection)
213

214
215
216
217
218
219
        l = A.container.locations
        if A not in l: l.append(A)
        l = B.container.locations
        if B not in l: l.append(B)
        

220
221
    # def _find_connections(self, loc):
    #     return [c for c in self.connections if c.A == loc or c.B == loc]
222
223

class SegmentParticle(PointParticle):
cmaffeo2's avatar
cmaffeo2 committed
224
    def __init__(self, type_, position, name="A", **kwargs):
225
        self.name = name
226
        self.contour_position = None
cmaffeo2's avatar
cmaffeo2 committed
227
        PointParticle.__init__(self, type_, position, name=name, **kwargs)
228
229
        self.intrahelical_neighbors = []
        self.other_neighbors = []
cmaffeo2's avatar
cmaffeo2 committed
230
        self.locations = []
231

232
    def get_intrahelical_above(self, all_types=True):
233
        """ Returns bead directly above self """
234
        # assert( len(self.intrahelical_neighbors) <= 2 )
235
        for b in self.intrahelical_neighbors:
236
            if b.get_contour_position(self.parent, self.contour_position) > self.contour_position:
237
238
                if all_types or isinstance(b,type(self)):
                    return b
239

240
    def get_intrahelical_below(self, all_types=True):
241
        """ Returns bead directly below self """
242
        # assert( len(self.intrahelical_neighbors) <= 2 )
243
        for b in self.intrahelical_neighbors:
244
            if b.get_contour_position(self.parent, self.contour_position) < self.contour_position:
245
246
                if all_types or isinstance(b,type(self)):
                    return b
247

cmaffeo2's avatar
cmaffeo2 committed
248
    def _neighbor_should_be_added(self,b):
249
250
251
        if type(self.parent) != type(b.parent):
            return True

cmaffeo2's avatar
cmaffeo2 committed
252
        c1 = self.contour_position
253
        c2 = b.get_contour_position(self.parent,c1)
cmaffeo2's avatar
cmaffeo2 committed
254
255
        if c2 < c1:
            b0 = self.get_intrahelical_below()
256
        else:
cmaffeo2's avatar
cmaffeo2 committed
257
258
259
            b0 = self.get_intrahelical_above()

        if b0 is not None:            
260
            c0 = b0.get_contour_position(self.parent,c1)
cmaffeo2's avatar
cmaffeo2 committed
261
262
263
264
265
266
267
268
269
270
271
272
273
            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:
274
275
            # assert(len(b.intrahelical_neighbors) <= 1)
            # assert(len(self.intrahelical_neighbors) <= 1)
cmaffeo2's avatar
cmaffeo2 committed
276
277
            self.intrahelical_neighbors.append(b)
            b.intrahelical_neighbors.append(self)
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
303
304
    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
305
306
307
308
        """ Returns the "address" of the nucleotide relative to seg in
        nucleotides, taking the shortest (intrahelical) contour length route to seg
        """
        if seg == self.parent:
309
            pos = self.contour_position
cmaffeo2's avatar
cmaffeo2 committed
310
        else:
311
312
            pos = self.get_contour_position(seg,near_address)
        return seg.contour_to_nt_pos(pos)
cmaffeo2's avatar
cmaffeo2 committed
313

314
    def get_contour_position(self,seg, address = None):
cmaffeo2's avatar
cmaffeo2 committed
315
316
317
318
        """ 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
        """

319
320
321
        if seg == self.parent:
            return self.contour_position
        else:
cmaffeo2's avatar
cmaffeo2 committed
322
323
324
325
326
            cutoff = 30*3
            target_seg = seg

            ## depth-first search
            ## TODO cache distances to nearby locations?
cmaffeo2's avatar
cmaffeo2 committed
327
            def descend_search_tree(seg, contour_in_seg, distance=0, visited_segs=None):
cmaffeo2's avatar
cmaffeo2 committed
328
                nonlocal cutoff
cmaffeo2's avatar
cmaffeo2 committed
329
                if visited_segs is None: visited_segs = []
cmaffeo2's avatar
cmaffeo2 committed
330
331
332
333

                if seg == target_seg:
                    # pdb.set_trace()
                    ## Found a segment in our target
334
335
                    sign = 1 if contour_in_seg == 1 else -1
                    if sign == -1: assert( contour_in_seg == 0 )
cmaffeo2's avatar
cmaffeo2 committed
336
337
                    if distance < cutoff: # TODO: check if this does anything
                        cutoff = distance
338
                    return [[distance, contour_in_seg+sign*seg.nt_pos_to_contour(distance)]], [(seg, contour_in_seg, distance)]
cmaffeo2's avatar
cmaffeo2 committed
339
                if distance > cutoff:
340
                    return None,None
cmaffeo2's avatar
cmaffeo2 committed
341
342
                    
                ret_list = []
343
                hist_list = []
cmaffeo2's avatar
cmaffeo2 committed
344
                ## Find intrahelical locations in seg that we might pass through
345
346
347
348
349
                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
350
                    if B.container in visited_segs: continue
351
352
353
                    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
354
355
356
                                                   distance+dx, visited_segs + [seg] )
                    if results is not None:
                        ret_list.extend( results )
357
358
                        hist_list.extend( history )
                return ret_list,hist_list
cmaffeo2's avatar
cmaffeo2 committed
359

360
            results,history = descend_search_tree(self.parent, self.contour_position)
cmaffeo2's avatar
cmaffeo2 committed
361
362
            if results is None or len(results) == 0:
                raise Exception("Could not find location in segment") # TODO better error
363
364
365
366
            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
367
368
369
            # nt_pos = self.get_nt_position(seg)
            # return seg.nt_pos_to_contour(nt_pos)

370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
    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
394
395
396
397
398
399
400
401
402
403
404
405
    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
406
407
408
    def __repr__(self):
        return "<SegmentParticle {} on {}[{:.2f}]>".format( self.name, self.parent, self.contour_position)

409
410

## TODO break this class into smaller, better encapsulated pieces
411
412
413
414
415
416
417
418
419
420
421
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
422
423
424
425
426
    orientation_particle = ParticleType("O",
                                        diffusivity = 100,
                                        mass = 300,
                                        radius = 1,
                                    )
427

cmaffeo2's avatar
cmaffeo2 committed
428
    # orientation_bond = HarmonicBond(10,2)
429
    orientation_bond = HarmonicBond(30,1.5, rRange = (0,500) )
430
431
432
433
434
435
436

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

437
    def __init__(self, name, num_nt, 
cmaffeo2's avatar
cmaffeo2 committed
438
                 start_position = None,
439
                 end_position = None, 
cmaffeo2's avatar
cmaffeo2 committed
440
441
                 segment_model = None,
                 **kwargs):
442

cmaffeo2's avatar
cmaffeo2 committed
443
444
        if start_position is None: start_position = np.array((0,0,0))

cmaffeo2's avatar
cmaffeo2 committed
445
        Group.__init__(self, name, children=[], **kwargs)
446
        ConnectableElement.__init__(self, connection_locations=[], connections=[])
447

cmaffeo2's avatar
cmaffeo2 committed
448
449
450
        if 'segname' not in kwargs:
            self.segname = name
        # self.resname = name
cmaffeo2's avatar
cmaffeo2 committed
451
452
453
454
455
        self.start_orientation = None
        self.twist_per_nt = 0

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

456
457
458
        self._bead_model_generation = 0    # TODO: remove?
        self.segment_model = segment_model # TODO: remove?

459
460
461
462
        self.strand_pieces = dict()
        for d in ('fwd','rev'):
            self.strand_pieces[d] = []

463
        self.num_nt = int(num_nt)
464
        if end_position is None:
465
            end_position = np.array((0,0,self.distance_per_nt*num_nt)) + start_position
466
467
468
        self.start_position = start_position
        self.end_position = end_position

cmaffeo2's avatar
cmaffeo2 committed
469
470
471
472
        ## Used to assign cadnano names to beads
        self._generate_bead_callbacks = []
        self._generate_nucleotide_callbacks = []

473
        ## Set up interpolation for positions
474
475
476
477
        self._set_splines_from_ends()

        self.sequence = None

478
    def __repr__(self):
479
        return "<{} {}[{:d}]>".format( str(type(self)).split('.')[-1], self.name, self.num_nt )
480

cmaffeo2's avatar
cmaffeo2 committed
481
    def set_splines(self, contours, coords):
482
        tck, u = interpolate.splprep( coords.T, u=contours, s=0, k=1)
483
        self.position_spline_params = (tck,u)
484

485
486
    def set_orientation_splines(self, contours, quaternions):
        tck, u = interpolate.splprep( quaternions.T, u=contours, s=0, k=1)
487
488
489
490
491
492
        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)

493
494
495
496
    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):
497
            positions[i] = self.contour_to_position(c)
498
499
500
501
502
            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_

503
504
505
    def _get_location_positions(self):
        return [self.contour_to_nt_pos(l.address) for l in self.locations]

cmaffeo2's avatar
cmaffeo2 committed
506
    def insert_dna(self, at_nt: int, num_nt: int, seq=tuple()):
507
508
509
510
511
512
513
514
515
516
517
518
        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
519
520
        ## TODO: handle sequence

521
522
523
524
525
        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)

526
    def remove_dna(self, first_nt:int, last_nt:int, remove_locations:bool = False):
cmaffeo2's avatar
cmaffeo2 committed
527
        """ Removes nucleotides between first_nt and last_nt, inclusive """
528
529
530
531
532
533
534
535
536
537
538
539
540
        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

541
542
543
544
545
546
547
548
549
        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

550
551
552
        first_nt = np.around(first_nt)
        last_nt = np.around(last_nt)

cmaffeo2's avatar
cmaffeo2 committed
553
554
555
        removed_nt = last_nt-first_nt+1
        num_nt = self.num_nt-removed_nt

556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
        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
572
573
574
575
        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 )
576

577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
        ## 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
598
        self.num_nt = num_nt
599

600
601
602
603
    def delete(self):
        for c,loc,other in self.get_connections_and_locations():
            c.delete()
        self.parent.segments.remove(self)
604

605
606
607
608
609
610
611
612
613
614
615
616
    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

617
    def translate(self, translation_vector, position_filter=None, contour_filter=None):
618
619
620
621
622
        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
623
        if len(ids) == 0: return
624
625
626

        ## Translate
        r[ids,:] = r[ids,:] + dr[np.newaxis,:]
627
        self.set_splines(u,r)
628
629
630
631
632
633

    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
634
        if len(ids) == 0: return
635
636
637

        if about is None:
            ## TODO: do this more efficiently
638
            r[ids,:] = np.array([rotation_matrix.dot(r[i,:]) for i in ids])
639
640
641
        else:
            dr = np.array(about)
            ## TODO: do this more efficiently
642
            r[ids,:] = np.array([rotation_matrix.dot(r[i,:]-dr) + dr for i in ids])
643

644
        self.set_splines(u,r)
645
646
647
648

        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
649
            orientations = np.array([self.contour_to_orientation(v) for v in u])
650
651
            for i in ids:
                orientations[i,:] = rotation_matrix.dot(orientations[i])
652
            quats = np.array([quaternion_from_matrix(o) for o in orientations])
653
            self.set_orientation_splines(u, quats)
654

655
    def _set_splines_from_ends(self, resolution=4):
656
        self.quaternion_spline_params = None
657
658
659
660
661
662
        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)
663

664
665
666
    def clear_all(self):
        Group.clear_all(self)  # TODO: use super?
        self.beads = []
667
668
669
670
671
        # for c,loc,other in self.get_connections_and_locations():
        #     loc.particle = None
        #     other.particle = None
        for l in self.locations:
            l.particle = None
672

673
    def contour_to_nt_pos(self, contour_pos, round_nt=False):
674
        nt = contour_pos*(self.num_nt) - 0.5
675
        if round_nt:
cmaffeo2's avatar
cmaffeo2 committed
676
            assert( np.isclose(np.around(nt),nt) )
677
678
679
            nt = np.around(nt)
        return nt

680
    def nt_pos_to_contour(self,nt_pos):
681
        return (nt_pos+0.5)/(self.num_nt)
682

683
    def contour_to_position(self,s):
684
        p = interpolate.splev( s, self.position_spline_params[0] )
685
686
687
688
        if len(p) > 1: p = np.array(p).T
        return p

    def contour_to_tangent(self,s):
689
        t = interpolate.splev( s, self.position_spline_params[0], der=1 )
690
691
        t = (t / np.linalg.norm(t,axis=0))
        return t.T
692
693
694
        

    def contour_to_orientation(self,s):
695
        assert( isinstance(s,float) or isinstance(s,int) or len(s) == 1 )   # TODO make vectorized version
696

697
698
699
700
701
        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)
702
            zAxis = np.array((0,0,1))
703
704

            if rotAxisL > 0.001:
705
706
707
                theta = np.arcsin(rotAxisL) * 180/np.pi
                if axis.dot(zAxis) < 0: theta = 180-theta
                orientation0 = rotationAboutAxis( rotAxis/rotAxisL, theta, normalizeAxis=False ).T
708
            else:
709
710
                orientation0 = np.eye(3) if axis.dot(zAxis) > 0 else \
                               rotationAboutAxis( np.array((1,0,0)), 180, normalizeAxis=False )
711
712
713
            if self.start_orientation is not None:
                orientation0 = orientation0.dot(self.start_orientation)

714
715
716
            orientation = rotationAboutAxis( axis, self.twist_per_nt*self.contour_to_nt_pos(s), normalizeAxis=False )
            orientation = orientation.dot(orientation0)
        else:
717
            q = interpolate.splev( s, self.quaternion_spline_params[0] )
718
719
            if len(q) > 1: q = np.array(q).T # TODO: is this needed?
            orientation = quaternion_to_matrix(q)
720

721
        return orientation
722

723
724
725
726
727
728
729
    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:
730
731
732
733
734
            try:
                c,A,B = self.get_contour_sorted_connections_and_locations(type_='intrahelical')[0 if nt_pos < 0 else -1]
            except:
                return None

735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
            ## 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
757
    def get_contour_sorted_connections_and_locations(self,type_):
cmaffeo2's avatar
cmaffeo2 committed
758
        sort_fn = lambda c: c[1].address
cmaffeo2's avatar
cmaffeo2 committed
759
        cl = self.get_connections_and_locations(type_)
cmaffeo2's avatar
cmaffeo2 committed
760
        return sorted(cl, key=sort_fn)
761
762
763
    
    def randomize_unset_sequence(self):
        bases = list(seqComplement.keys())
764
        # bases = ['T']        ## FOR DEBUG
765
        if self.sequence is None:
766
            self.sequence = [random.choice(bases) for i in range(self.num_nt)]
767
        else:
768
            assert(len(self.sequence) == self.num_nt) # TODO move
769
770
771
            for i in range(len(self.sequence)):
                if self.sequence[i] is None:
                    self.sequence[i] = random.choice(bases)
772

cmaffeo2's avatar
cmaffeo2 committed
773
774
775
    def _get_num_beads(self, max_basepairs_per_bead, max_nucleotides_per_bead ):
        raise NotImplementedError

776
    def _generate_one_bead(self, contour_position, nts):
777
778
        raise NotImplementedError

cmaffeo2's avatar
cmaffeo2 committed
779
    def _generate_atomic_nucleotide(self, contour_position, is_fwd, seq, scale, strand_segment):
cmaffeo2's avatar
cmaffeo2 committed
780
781
782
783
784
        """ 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)
785
        orientation = self.contour_to_orientation(contour_position)
cmaffeo2's avatar
cmaffeo2 committed
786

787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
        """ 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
805
        key = seq
806
807
        nt_dict = canonicalNtFwd if is_fwd else canonicalNtRev

808
        atoms = nt_dict[ key ].generate() # TODO: clone?
cmaffeo2's avatar
cmaffeo2 committed
809
        atoms.orientation = orientation.dot(atoms.orientation)
810
811
812
813
        if isinstance(self, SingleStrandedSegment):
            if scale is not None and scale != 1:
                for a in atoms:
                    a.position = scale*a.position
814
            atoms.position = pos - atoms.atoms_by_name["C1'"].collapsedPosition()
815
816
817
818
819
820
821
822
823
        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
824
                    else:
825
                        a.fixed = 1
826
            atoms.position = pos
cmaffeo2's avatar
cmaffeo2 committed
827
828
829
830
831
832
        
        atoms.contour_position = contour_position
        strand_segment.add(atoms)

        for callback in self._generate_nucleotide_callbacks:
            callback(atoms)
cmaffeo2's avatar
cmaffeo2 committed
833
834

        return atoms
835

836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
    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


858
859
    def add_location(self, nt, type_, on_fwd_strand=True):
        ## Create location if needed, add to segment
860
        c = self.nt_pos_to_contour(nt)
861
862
863
864
865
866
        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?
867
868

    def add_nick(self, nt, on_fwd_strand=True):
869
870
871
872
873
874
875
        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)

876

877
    def add_5prime(self, nt, on_fwd_strand=True):
878
879
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
880
        self.add_location(nt,"5prime",on_fwd_strand)
881
882

    def add_3prime(self, nt, on_fwd_strand=True):
883
884
        if isinstance(self,SingleStrandedSegment):
            on_fwd_strand = True
885
        self.add_location(nt,"3prime",on_fwd_strand)
886

887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
    ## 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 )

911
    def get_3prime_locations(self):
cmaffeo2's avatar
cmaffeo2 committed
912
        return sorted(self.get_locations("3prime"),key=lambda x: x.address)
913
    
cmaffeo2's avatar
cmaffeo2 committed
914
    def get_5prime_locations(self):
915
        ## TODO? ensure that data is consistent before _build_model calls
cmaffeo2's avatar
cmaffeo2 committed
916
        return sorted(self.get_locations("5prime"),key=lambda x: x.address)
cmaffeo2's avatar
cmaffeo2 committed
917

918
919
920
921
922
923
924
925
926
927
928
    def get_blunt_DNA_ends(self):
        if isinstance(self, SingleStrandedSegment): return []
        ret = []
        cl = self.get_connections_and_locations("intrahelical")
        if not any([c[1].address == 0 for c in cl]):
            ret.append((self.start5,self.start3,-1))
        if not any([c[1].address == 1 for c in cl]):
            ret.append((self.end5,self.end3,1))
        # return [c for c in cl if c[1].address == 0 or c[1].address == 1]
        return ret

929
    def iterate_connections_and_locations(self, reverse=False):
cmaffeo2's avatar
cmaffeo2 committed
930
931
        ## connections to other segments
        cl = self.get_contour_sorted_connections_and_locations()
932
        if reverse:
cmaffeo2's avatar
cmaffeo2 committed
933
            cl = cl[::-1]
934
935
936
            
        for c in cl:
            yield c
cmaffeo2's avatar
cmaffeo2 committed
937

938
939
940
941
942
943
944
945
946
947
948
    ## 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):
949
950
                if value >= l and value <= h:
                    raise Exception("Strand piece already exists! DNA may be circular.")
951
952
953
954
955
956

        ## 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)

957
    ## TODO rename
958
    def get_strand_segment(self, nt_pos, is_fwd, move_at_least=0.5):
959
        """ Walks through locations, checking for crossovers """
960
961
962
963
        # if self.name in ("6-1","1-1"):
        #     import pdb
        #     pdb.set_trace()
        move_at_least = 0
964
965

        ## Iterate through locations
cmaffeo2's avatar
cmaffeo2 committed
966
        # locations = sorted(self.locations, key=lambda l:(l.address,not l.on_fwd_strand), reverse=(not is_fwd))
967
968
969
970
971
972
        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))
973
974
        # print(locations)

975
        for l in locations:
cmaffeo2's avatar
cmaffeo2 committed
976
977
978
979
            # TODOTODO probably okay
            if l.address == 0:
                pos = 0.0
            elif l.address == 1:
980
                pos = self.num_nt-1
cmaffeo2's avatar
cmaffeo2 committed
981
982
            else:
                pos = self.contour_to_nt_pos(l.address, round_nt=True)
983
984
985

            ## DEBUG

cmaffeo2's avatar
cmaffeo2 committed
986

987
            ## Skip locations encountered before our strand
988
989
990
991
992
993
994
995
            # 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
996
997

            ## Stop if we found the 3prime end
998
            if l.on_fwd_strand == is_fwd and l.type_ == "3prime" and l.connection is None:
999
1000
                if _DEBUG_TRACE:
                    print("  found end at",l)
1001
                return pos, None, None, None, None
1002
1003
1004
1005
1006
1007
1008
1009

            ## 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:
1010
1011
1012
                if _DEBUG_TRACE:
                    print("  passing through",l)
                    print("from {}, connection {} to {}".format(nt_pos,l,B))
1013
                Bpos = B.get_nt_pos()
1014
                return pos, B.container, Bpos, B.on_fwd_strand, 0.5
1015
1016
1017
                
            ## Stop at other strand crossovers so basepairs line up
            elif c.type_ == "crossover":
1018
                if nt_pos == pos: continue
1019
1020
                if _DEBUG_TRACE:
                    print("  pausing at",l)
1021
                return pos, l.container, pos+(2*is_fwd-1), is_fwd, 0
1022
1023
1024
1025
1026
1027

        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

cmaffeo2's avatar