libprosic 0.4.0

A library for calling of genomic variants using a latent variable model.
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
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
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
use itertools::{Itertools};
use ordered_float::NotNaN;
use bio::stats::LogProb;
use statrs::function::beta::ln_beta;
use statrs::function::factorial::ln_binomial;

use model::{Variant, ContinuousAlleleFreqs, DiscreteAlleleFreqs, AlleleFreq};

use priors::PairModel;

/// Prior model for a Single Cell against a Bulk background from the same individual (optimally the
/// same cell type). It uses the ploidy of the organism as well as a WGA method specific single cell
/// model (so far only for MDA) to account for differential allelic amplification.
/// TODO: * use the general level of heterozygosity through the InfiniteSitesNeutralEvolutionModel as
///         a prior? "The prior probability for a germline allele frequency theta_g (e.g. 0.0, 0.5 or 1.0 for the diploid case) in the bulk background can be calculated with an `InfiniteSitesNeutralVariationModel`. This is valid since clonal variants come from the last common ancestor and analogously to tumor evolution in the Williams model, we can assume neutral mutations (no genetic drift, no selection) and thus no change of allele frequencies in cell divisions that do not introduce new mutations. The `InfiniteSitesNeutralVariationModel` requires the ploidy and the level of heterozygosity."
///       * use the somatic mutation rate per effective cell division? ("The somatic mutation rate per effective cell division in the bulk is the quotient mu/beta, with mu being the somatic mutation rate and beta being the fraction of effective cell divisions (i.e. where both daugther cells survive and form a lineage). Alone, these parameters are not easily obtained. However, assuming mostly neutral mutations, mu/beta can be estimated from SNV calls with a low frequency in the bulk sample, analogous to the tumour sample in Williams et al. (2016). It is the slope of the linear model `y = mu/beta * (x -  1 / fmax)`, with `x` being the reciprocal of the observed allele frequencies and y being the number of observed mutations corresponding to each frequency (see: Williams MJ, Werner B, Barnes CP, Graham TA, Sottoriva A. Identification of neutral tumor evolution across cancer types. Nat Genet. 2016;48: 238–244. doi:10.1038/ng.3489). Based on the Williams model, the tail probability of a somatic allele frequency F > f can be expressed as `Pr(F > f) = M(f) / n = mu/beta (1 / f - 1 / fmax) / n`, with `n` being the size of the genome and `fmax` the expected allele frequency of clonal variants at the beginning of tumor history, overall somatic history in our case. From this, we can obtain the cumulative distribution function as `Pr(F <= f) = 1 - Pr(F > f)`. Consequently, the density becomes the first derivative, i.e. `Pr(F = f) = - M(f)' / n = mu/beta * 1/n * 1/f²` for `f>=fmin`, with `fmin = sqrt(mu/beta * 1/n)`."
pub struct SingleCellBulkModel {
    allele_freqs_single: DiscreteAlleleFreqs,
    allele_freqs_bulk: ContinuousAlleleFreqs,
    n_bulk_min: usize,
    n_bulk_max: usize
}

impl SingleCellBulkModel {
    /// Create new model.
    ///
    /// # Arguments
    ///
    /// * `ploidy` - the ploidy in the single cell sample (e.g. 2 for diploid)
    /// * `n_bulk_min` - minimum number of discrete frequencies n_bulk_min+1 to evaluate over interval [0,1] for bulk sample; if coverage is above this, actual read counts are used to test all possible discrete allele frequencies
    /// * `n_bulk_max` - maximum number of discrete frequencies n_bulk_min+1 to evaluate over interval [0,1] for bulk sample; if coverage is below this, actual read counts are used to test all possible discrete allele frequencies
    pub fn new(ploidy: u32, n_bulk_min: usize, n_bulk_max: usize) -> Self {
        let allele_freqs = (0..ploidy + 1).map(|m| AlleleFreq(m as f64 / ploidy as f64)).collect_vec();
        SingleCellBulkModel {
            allele_freqs_single: allele_freqs,
            allele_freqs_bulk: ContinuousAlleleFreqs::inclusive( 0.0..1.0 ),
            n_bulk_min: n_bulk_min, // TODO: implement initialization via command line arguments in ProSolo, i.e. via n_bulk_per_event_min and iteration over events defined in ProSolo
            n_bulk_max: n_bulk_max // TODO: implement initialization via command line arguments in ProSolo
        }
    }

    // Lodato et al. 2015, Science, Supplementary Information, pages 8f and Fig. S5 (A, C, E)
    // TODO: allow for non-default Lodato model parameters, e.g. learned from the data at hand
    // TODO: allow for non-Lodato models
    fn prob_rho(
        &self,
        af_single_underlying: &f64,
        n_obs: &usize,
        k: &usize
    ) -> LogProb
    {
        let binomial_coeff = ln_binomial(*n_obs as u64,*k as u64);

        match *af_single_underlying {
            // model for hom ref sites
            0.0 => {
                let alpha = |cov| {
                    -0.000027183 * cov as f64 + 0.068567471
                };
                let beta = |cov| {
                    0.007454388 * cov as f64 + 2.367486659
                };
                let a = alpha(*n_obs);
                let b = beta(*n_obs);

                LogProb(
                    binomial_coeff +
                        ln_beta((*k as f64) + a,  (*n_obs) as f64 - (*k as f64) + b) - ln_beta(a,b)
                )
            },
            // model for heterozygous sites
            0.5 => {
                let weight = |cov| {
                    0.000548761 * cov as f64 + 0.540396786
                };
                let alpha_1 = |cov| {
                    0.057378844 * cov as f64 + 0.669733191
                };
                let alpha_2 = |cov| {
                    0.003233912 * cov as f64 + 0.399261625
                };

                let a1 = alpha_1(*n_obs);
                let a2 = alpha_2(*n_obs);
                let w = weight(*n_obs);

                // Lodato et al. 2015, Science, Supplementary Information, pages 8f and Fig. S5 (A, C, E)
                LogProb(
                    w.ln() + binomial_coeff +
                        ln_beta((*k as f64) + a1,  (*n_obs) as f64 - (*k as f64) + a1) - ln_beta(a1,a1)
                ).ln_add_exp(
                    LogProb(
                        ((-w).ln_1p()) + binomial_coeff +
                            ln_beta((*k as f64) + a2, (*n_obs) as f64 - (*k as f64) + a2) - ln_beta(a2,a2) )
                )
            },
            // model for hom alt sites (hom ref density mirrored)
            1.0 => {
                let alpha = |cov| {
                    0.007454388 * cov as f64 + 2.367486659
                };
                let beta = |cov| {
                    -0.000027183 * cov as f64 + 0.068567471
                };
                let a = alpha(*n_obs);
                let b = beta(*n_obs);

                LogProb(
                    binomial_coeff +
                        ln_beta((*k as f64) + a,  (*n_obs) as f64 - (*k as f64) + b) - ln_beta(a,b)
                )
            },
            _ => panic!("SingleCellBulkModel is currently only implemented for the diploid case with allele frequencies 0.0, 0.5 and 1.0.")
        }
    }

    fn prior_bulk(&self, _: AlleleFreq, _: &Variant) -> LogProb {
        // TODO: stick in meaningful prior for the bulk sample, derivative of InfiniteSitesNeutralVariationModel?
        LogProb::ln_one()
    }

    fn prior_single(&self, _: AlleleFreq, _: &Variant) -> LogProb {
        // TODO: stick in meaningful prior for the single cell sample, e.g. P(theta_b > 0 | bulk_pileup)
        LogProb::ln_one()
    }

    /// Function to cap the use of the single cell amplification bias model at a coverage of 100,
    /// as the Lodato et al. model was fit with coverages capped at 60 and starts behaving
    /// weirdly above 100
    fn cap_n_s(&self, n: usize) -> usize {
    // TODO: make this optional and dependent on the usage of the Lodato model with their params
        if n > 100 {
            100
        } else {
           n
        }
    }

    /// adjust n to use for bulk coverage at current site, using application-set min and max values
    fn adjust_n_b(&self, n: usize) -> usize {
        if n <= self.n_bulk_min {
            self.n_bulk_min
        } else if n >= self.n_bulk_max {
            self.n_bulk_max
        } else {
            n
        }
    }

    /// determine values of k to use with current n and given allele frequency range
    fn determine_k_b(&self, af: &ContinuousAlleleFreqs, n: usize) -> (usize,usize) {
        let k_start = if af.left_exclusive {
            (*af.start * n as f64).floor() + 1.0
        } else {
            (*af.start * n as f64).ceil()
        };
        let k_end = if af.right_exclusive {
            (*af.end * n as f64).ceil()
        } else {
            (*af.end * n as f64).floor() + 1.0
        };
        assert!(k_end > k_start, "One of the bulk event ranges defined by your application is too small to be covered with the current minimum number of discrete bulk allele frequencies requested (k_end !> k_start).");
        (k_start as usize, k_end as usize )
    }
}


impl PairModel<DiscreteAlleleFreqs, ContinuousAlleleFreqs> for SingleCellBulkModel {

    fn prior_prob(
        &self,
        af_single: AlleleFreq,
        af_bulk: AlleleFreq,
        variant: &Variant
    ) -> LogProb {
        self.prior_single(af_single, variant) + self.prior_bulk(af_bulk, variant)
    }

    fn joint_prob<L, O>(
        &self,
        af_single: &DiscreteAlleleFreqs,
        af_bulk: &ContinuousAlleleFreqs,
        likelihood_single_distorted: &L,
        likelihood_bulk: &O,
        variant: &Variant,
        n_obs_single: usize,
        n_obs_bulk: usize
    ) -> LogProb where
        L: Fn(AlleleFreq, Option<AlleleFreq>) -> LogProb,
        O: Fn(AlleleFreq, Option<AlleleFreq>) -> LogProb
    {
        let n_single = self.cap_n_s(n_obs_single);
        let k_single = 0..n_single + 1;

        let n_bulk = self.adjust_n_b(n_obs_bulk);
        let (k_start, k_end) = self.determine_k_b(af_bulk, n_bulk);
        let k_bulk = k_start..k_end;
        // sum up all possible discrete bulk allele frequencies with current number of observations
        let p_bulk = LogProb::ln_sum_exp(&k_bulk.map(|k_b| {
            let af_bulk = AlleleFreq(k_b as f64/n_bulk as f64);
            let p: LogProb;
            if n_obs_bulk == 0 { // speedup for zero coverage sites
                p = self.prior_bulk(af_bulk, variant)
            } else {
                p = self.prior_bulk(af_bulk, variant) + likelihood_bulk(af_bulk, None)
            }
            p
        }).collect_vec() );

/*
        // do Simpson's integration instead of point-wise evaluation above
        let density = |af_bulk| {
            let af_bulk = AlleleFreq(af_bulk);
            self.prior_bulk(af_bulk, variant) +
            likelihood_bulk(af_bulk, None)
        };
        let grid_points = k_end - k_start;
        println!("bulk grid_points: {}", grid_points);

        let p_bulk = if af_bulk.start == af_bulk.end {
                density(*af_bulk.start)
            } else {
                LogProb::ln_simpsons_integrate_exp(
                    &density,
                    *af_bulk.start,
                    *af_bulk.end,
                    if grid_points % 2 == 1 {
                        grid_points
                    } else {
                        grid_points + 1
                    })
            };
*/

        // go through all possible underlying single cell allele frequencies
        let prob = LogProb::ln_sum_exp(&af_single.iter().map(|&af_single| {
            let p_single =
                    LogProb::ln_sum_exp(&k_single.clone().map(|k_s| { // sum up all possible discrete single cell allele frequencies with current number of observations
                        if n_single == 0 {
                            self.prior_single(af_single, variant)
                        } else {
                            let af_single_distorted = AlleleFreq(k_s as f64/n_single as f64);
                            self.prior_single(af_single, variant) +
                            likelihood_single_distorted(af_single_distorted, None) +
                            self.prob_rho(&af_single, &n_single, &k_s)
                        }
                    }).collect_vec());
            let prob = p_bulk + p_single;

            prob
        }).collect_vec());

        prob
    }

    fn marginal_prob<L, O>(
        &self,
        likelihood_single_distorted: &L,
        likelihood_bulk: &O,
        variant: &Variant,
        n_obs_single: usize,
        n_obs_bulk: usize
    ) -> LogProb where
        L: Fn(AlleleFreq, Option<AlleleFreq>) -> LogProb,
        O: Fn(AlleleFreq, Option<AlleleFreq>) -> LogProb
    {
        self.joint_prob(
            self.allele_freqs().0,
            self.allele_freqs().1,
            likelihood_single_distorted,
            likelihood_bulk,
            variant,
            n_obs_single,
            n_obs_bulk
        )
    }

    fn map<L, O>(
        &self,
        likelihood_single_distorted: &L,
        likelihood_bulk: &O,
        variant: &Variant,
        n_obs_single: usize,
        n_obs_bulk: usize
    ) -> (AlleleFreq, AlleleFreq) where
        L: Fn(AlleleFreq, Option<AlleleFreq>) -> LogProb,
        O: Fn(AlleleFreq, Option<AlleleFreq>) -> LogProb
    {
        let n_single = self.cap_n_s(n_obs_single);
        let k_single = 0..n_single + 1;

        let (_, map_single) = self.allele_freqs().0.iter().minmax_by_key(
            |&&af_single| {
                let p_single =
                    LogProb::ln_sum_exp(&k_single.clone().map(|k_s| { // sum up all possible discrete single cell allele frequencies with current number of observations
                        if n_single == 0 { // avoid division by zero, and speedup
                            self.prior_single(af_single, variant)
                        } else {
                            let af_single_distorted = AlleleFreq(k_s as f64/n_single as f64);
                            self.prior_single(af_single, variant) +
                            likelihood_single_distorted(af_single_distorted, None) +
                            self.prob_rho(&af_single, &n_single, &k_s)
                        }
                    }).collect_vec());
                NotNaN::new(*p_single).expect("posterior probability is NaN")
            }
        ).into_option().expect("prior has empty allele frequency spectrum");

        let n_bulk = self.adjust_n_b(n_obs_bulk);
        let k_bulk = 0..n_bulk + 1;
        let afs_bulk = k_bulk.map(|k| AlleleFreq(k as f64/n_bulk as f64));
        let (_, map_bulk) = afs_bulk.minmax_by_key(
            |af_bulk| {
                let p_bulk = self.prior_bulk(*af_bulk, variant) + likelihood_bulk(*af_bulk, None);
                NotNaN::new(*p_bulk).expect("posterior probability is NaN")
            }
        ).into_option().expect("prior has empty allele frequency spectrum");

        (*map_single, map_bulk)
    }

    fn allele_freqs(&self) -> (&DiscreteAlleleFreqs, &ContinuousAlleleFreqs) {
        (&self.allele_freqs_single, &self.allele_freqs_bulk)
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use bio::stats::LogProb;
    use model::{ContinuousAlleleFreqs, AlleleFreq, AlleleFreqs, likelihood, Variant, PairPileup, priors};
    use model::sample::{Observation, Evidence};

    #[test]
    fn test_prob_rho() {
        let model = SingleCellBulkModel::new(2,5,100);
        // all expected results calculated with implementation in R version 3.3.3, using
        // dbetabinom.ab() from the R package VGAM_1.0-2

        // HOM REF model
        let results_5_hom_ref = [0.917057691405, 0.048991452054, 0.019369602690, 0.009095775973, 0.004098640948, 0.001386836930];
        let results_60_hom_ref = [  0.803320245461, 0.052192823008, 0.027012354914, 0.018046293954, 0.013409772796, 0.010564967885,
                                    0.008637021122, 0.007242360970, 0.006185855926, 0.005357571136, 0.004690762755, 0.004142520913,
                                    0.003683979192, 0.003294990190, 0.002961056264, 0.002671473498, 0.002418163441, 0.002194912251,
                                    0.001996860416, 0.001820151668, 0.001661685948, 0.001518942088, 0.001389848262, 0.001272685809,
                                    0.001166016809, 0.001068628819, 0.000979492206, 0.000897726836, 0.000822575822, 0.000753384624,
                                    0.000689584283, 0.000630677854, 0.000576229360, 0.000525854718, 0.000479214255, 0.000436006489,
                                    0.000395962936, 0.000358843750, 0.000324434041, 0.000292540765, 0.000262990063, 0.000235625005,
                                    0.000210303643, 0.000186897346, 0.000165289364, 0.000145373587, 0.000127053478, 0.000110241151,
                                    0.000094856580, 0.000080826922, 0.000068085955, 0.000056573603, 0.000046235582, 0.000037023138,
                                    0.000028892926, 0.000021807042, 0.000015733297, 0.000010645858, 0.000006526566, 0.000003367639, 0.000001177820];
        // HET model
        let results_5_het = [0.215810926125, 0.147529824097, 0.136659249777, 0.136659249777, 0.147529824097, 0.215810926125];
        let results_60_het = [  0.022964691724, 0.013791217110, 0.011203808536, 0.009976873288, 0.009344532591, 0.009066573603,
                                0.009039972286, 0.009210742960, 0.009545572404, 0.010020673323, 0.010616821444, 0.011316971031,
                                0.012105068132, 0.012965466790, 0.013882667547, 0.014841235701, 0.015825822418, 0.016821244988,
                                0.017812600331, 0.018785395739, 0.019725686644, 0.020620214709, 0.021456541698, 0.022223175971,
                                0.022909689396, 0.023506823069, 0.024006580679, 0.024402308658, 0.024688762485, 0.024862158676,
                                0.024920212140, 0.024862158676, 0.024688762485, 0.024402308658, 0.024006580679, 0.023506823069,
                                0.022909689396, 0.022223175971, 0.021456541698, 0.020620214709, 0.019725686644, 0.018785395739,
                                0.017812600331, 0.016821244988, 0.015825822418, 0.014841235701, 0.013882667547, 0.012965466790,
                                0.012105068132, 0.011316971031, 0.010616821444, 0.010020673323, 0.009545572404, 0.009210742960,
                                0.009039972286, 0.009066573603, 0.009344532591, 0.009976873288, 0.011203808536, 0.013791217110, 0.022964691724];
        // HOM ALT model
        let results_5_hom_alt = [0.001386836930, 0.004098640948, 0.009095775973, 0.019369602690, 0.048991452054, 0.917057691405];
        let results_60_hom_alt = [  0.000001177820, 0.000003367639, 0.000006526566, 0.000010645858, 0.000015733297, 0.000021807042,
                                    0.000028892926, 0.000037023138, 0.000046235582, 0.000056573603, 0.000068085955, 0.000080826922,
                                    0.000094856580, 0.000110241151, 0.000127053478, 0.000145373587, 0.000165289364, 0.000186897346,
                                    0.000210303643, 0.000235625005, 0.000262990063, 0.000292540765, 0.000324434041, 0.000358843750,
                                    0.000395962936, 0.000436006489, 0.000479214255, 0.000525854718, 0.000576229360, 0.000630677854,
                                    0.000689584283, 0.000753384624, 0.000822575822, 0.000897726836, 0.000979492206, 0.001068628819,
                                    0.001166016809, 0.001272685809, 0.001389848262, 0.001518942088, 0.001661685948, 0.001820151668,
                                    0.001996860416, 0.002194912251, 0.002418163441, 0.002671473498, 0.002961056264, 0.003294990190,
                                    0.003683979192, 0.004142520913, 0.004690762755, 0.005357571136, 0.006185855926, 0.007242360970,
                                    0.008637021122, 0.010564967885, 0.013409772796, 0.018046293954, 0.027012354914, 0.052192823008, 0.803320245461];
        // test all models
        for k in 0..5+1 {
            assert_relative_eq!( model.prob_rho(&AlleleFreq(0.0), &5, &(k as usize)).exp(),
                                    results_5_hom_ref[k] as f64, max_relative = 1.0,
                                    epsilon = 0.000000000001);
            assert_relative_eq!( model.prob_rho(&AlleleFreq(0.5), &5, &(k as usize)).exp(),
                                    results_5_het[k] as f64, max_relative = 1.0,
                                    epsilon = 0.000000000001);
            assert_relative_eq!( model.prob_rho(&AlleleFreq(1.0), &5, &(k as usize)).exp(),
                                    results_5_hom_alt[k] as f64, max_relative = 1.0,
                                    epsilon = 0.000000000001);
        }
        for k in 0..60+1 {
            assert_relative_eq!( model.prob_rho(&AlleleFreq(0.0), &60, &(k as usize)).exp(),
                                    results_60_hom_ref[k] as f64, max_relative = 1.0,
                                    epsilon = 0.000000000001);
            assert_relative_eq!( model.prob_rho(&AlleleFreq(0.5), &60, &(k as usize)).exp(),
                                    results_60_het[k] as f64, max_relative = 1.0,
                                    epsilon = 0.000000000001);
            assert_relative_eq!( model.prob_rho(&AlleleFreq(1.0), &60, &(k as usize)).exp(),
                                    results_60_hom_alt[k] as f64, max_relative = 1.0,
                                    epsilon = 0.000000000001);
        }
    }

    fn create_obs_vector(
        n_obs_ref: usize,
        n_obs_alt: usize
    ) -> Vec<Observation> {
        let obs_ref_abs = Observation {
            prob_mapping: LogProb::ln_one(),
            //prob_mapping: LogProb(0.9f64.ln()),
            prob_alt: LogProb::ln_zero(),
            prob_ref: LogProb::ln_one(),
            prob_mismapped: LogProb::ln_one(),
            evidence: Evidence::dummy_alignment()
        };
        let obs_alt_abs = Observation {
            prob_mapping: LogProb::ln_one(),
            //prob_mapping: LogProb(0.9f64.ln()),
            prob_alt: LogProb::ln_one(),
            prob_ref: LogProb::ln_zero(),
            prob_mismapped: LogProb::ln_one(),
            evidence: Evidence::dummy_alignment()
        };

        let mut obs = Vec::new();
        for _ in 0..n_obs_ref {
            obs.push(obs_ref_abs.clone());
        }
        for _ in 0..n_obs_alt {
            obs.push(obs_alt_abs.clone());
        }
        obs
    }

    fn create_test_snv_pileup<'a, A: AlleleFreqs, B: AlleleFreqs, P: priors::PairModel<A, B>>(
        prior_model: &'a P,
        s_obs: &Vec<Observation>,
        b_obs: &Vec<Observation>
    ) -> PairPileup<'a, A, B, P> {
        let variant = Variant::SNV(b'T');
        let single_sample_model = likelihood::LatentVariableModel::with_single_sample();
        let bulk_sample_model = likelihood::LatentVariableModel::with_single_sample();
        PairPileup::new(
            s_obs.clone(),
            b_obs.clone(),
            variant,
            prior_model,
            single_sample_model,
            bulk_sample_model,
        )
    }

    fn bulk_test_ranges_sum<'a, A: AlleleFreqs, B: AlleleFreqs, P: priors::PairModel<A, B>>(
        model: &'a P,
        s_obs: &Vec<Observation>,
        afs_s: &A,
        b_obs: &Vec<Observation>,
        rs_b: Vec<&B>
    ) {
        println!("SCBM bulk ranges discretization sum test with:");
        println!("  * {} bulk ranges", rs_b.len() );
        println!("  * bulk Observations: {:?}", b_obs );
        let pileup = create_test_snv_pileup(model, s_obs, b_obs);
        let p_j: Vec<LogProb> = rs_b.iter().map(|r| pileup.joint_prob(afs_s, r)).collect();
        for (i, r) in rs_b.iter().enumerate() {
            println!("  Bulk range: {:?}; pileup.joint_prob() = {}", r, p_j[i].exp() );
        }
        let p_m = pileup.marginal_prob();
        println!("  pileup.marginal_prob(): {}", p_m.exp() );
        let p_sum = p_j.iter().fold( LogProb::ln_zero(), |sum, p| sum.ln_add_exp(p - p_m) );
        assert_relative_eq!( p_sum.exp(), 1.0, epsilon = 0.0000000001 );
    }

    fn bulk_test_range_fraction<'a, A: AlleleFreqs, B: AlleleFreqs, P: priors::PairModel<A, B>>(
        model: &'a P,
        s_obs: &Vec<Observation>,
        afs_s: &A,
        b_obs: &Vec<Observation>,
        r_b: &B,
        fraction: f64
    ) {
        println!("SCBM bulk ranges discretization ratio test with bulk observations: {:?}.", b_obs);
        let pileup = create_test_snv_pileup(model, s_obs, b_obs);
        let p_j = pileup.joint_prob(afs_s, r_b);
        let p_m = pileup.marginal_prob();
        println!("  pileup.marginal_prob() = {}; bulk range: {:?}; pileup.joint_prob() = {}", p_m.exp(), r_b, p_j.exp() );
        assert_relative_eq!( ( p_j - p_m ).exp(), fraction, epsilon = 0.0000000001 );
    }

    // tests that bulk range discretization includes and excludes the right discrete values per
    // bulk frequency range (indirect through probabilities, as these bulk ranges aren't public)
    #[test]
    fn test_bulk_range_discretization() {
        let model_4 = SingleCellBulkModel::new(2,4,100);
        let model_5 = SingleCellBulkModel::new(2,5,100);
        let model_10 = SingleCellBulkModel::new(2,10,100);

        // static single cell setup
        let s_dummy_obs = create_obs_vector(10, 0);
        let af_single_all = vec![AlleleFreq(0.0), AlleleFreq(0.5), AlleleFreq(1.0)];

        // two-range setups
        let af_bulk_zero     = ContinuousAlleleFreqs::inclusive( 0.0..0.0 );
        let af_bulk_not_zero = ContinuousAlleleFreqs::left_exclusive( 0.0..1.0 );

        let af_bulk_not_one  = ContinuousAlleleFreqs::right_exclusive( 0.0..1.0 );
        let af_bulk_one      = ContinuousAlleleFreqs::inclusive( 1.0..1.0 );

        // three-range setups
        let af_bulk_below_quarter  = ContinuousAlleleFreqs::right_exclusive( 0.0..0.25 );
        let af_bulk_quarter        = ContinuousAlleleFreqs::inclusive( 0.25..0.25 );
        let af_bulk_above_quarter  = ContinuousAlleleFreqs::left_exclusive( 0.25..1.0 );

        let af_bulk_below_half     = ContinuousAlleleFreqs::right_exclusive( 0.0..0.5 );
        let af_bulk_half           = ContinuousAlleleFreqs::inclusive( 0.5..0.5 );
        let af_bulk_above_half     = ContinuousAlleleFreqs::left_exclusive( 0.5..1.0 );


        let b_1_ref = create_obs_vector(1, 0);
        bulk_test_ranges_sum(&model_10, &s_dummy_obs, &af_single_all,
                             &b_1_ref, vec![&af_bulk_zero, &af_bulk_not_zero]);
        bulk_test_ranges_sum(&model_5, &s_dummy_obs, &af_single_all,
                             &b_1_ref, vec![&af_bulk_not_one, &af_bulk_one]);
        bulk_test_range_fraction(&model_5, &s_dummy_obs, &af_single_all,
                             &b_1_ref, &af_bulk_zero, 0.33333333333333);
        bulk_test_ranges_sum(&model_4, &s_dummy_obs, &af_single_all,
                             &b_1_ref,
                             vec![&af_bulk_below_quarter, &af_bulk_quarter, &af_bulk_above_quarter]);
        bulk_test_ranges_sum(&model_10, &s_dummy_obs, &af_single_all,
                             &b_1_ref,
                             vec![&af_bulk_below_half, &af_bulk_half, &af_bulk_above_half]);

        let b_1_alt = create_obs_vector(0, 1);
        bulk_test_ranges_sum(&model_5, &s_dummy_obs, &af_single_all,
                             &b_1_alt, vec![&af_bulk_zero, &af_bulk_not_zero]);
        bulk_test_ranges_sum(&model_10, &s_dummy_obs, &af_single_all,
                             &b_1_alt, vec![&af_bulk_not_one, &af_bulk_one]);

        let b_2_ref = create_obs_vector(2, 0);
        bulk_test_ranges_sum(&model_10, &s_dummy_obs, &af_single_all,
                             &b_2_ref, vec![&af_bulk_zero, &af_bulk_not_zero]);
        bulk_test_ranges_sum(&model_4, &s_dummy_obs, &af_single_all,
                             &b_2_ref, vec![&af_bulk_not_one, &af_bulk_one]);

        let b_1_ref_1_alt = create_obs_vector(1, 1);
        bulk_test_ranges_sum(&model_4, &s_dummy_obs, &af_single_all,
                             &b_1_ref_1_alt, vec![&af_bulk_zero, &af_bulk_not_zero]);
        bulk_test_range_fraction(&model_4, &s_dummy_obs, &af_single_all,
                             &b_1_ref_1_alt, &af_bulk_half, 0.4);
        bulk_test_range_fraction(&model_4, &s_dummy_obs, &af_single_all,
                             &b_1_ref_1_alt, &af_bulk_zero, 0.0);
        bulk_test_ranges_sum(&model_10, &s_dummy_obs, &af_single_all,
                             &b_1_ref_1_alt, vec![&af_bulk_not_one, &af_bulk_one]);

        let b_10_ref = create_obs_vector(10, 0);
        bulk_test_ranges_sum(&model_4, &s_dummy_obs, &af_single_all,
                             &b_10_ref, vec![&af_bulk_zero, &af_bulk_not_zero]);
        bulk_test_range_fraction(&model_4, &s_dummy_obs, &af_single_all,
                             &b_10_ref, &af_bulk_zero, 0.67049555724866);
        bulk_test_range_fraction(&model_5, &s_dummy_obs, &af_single_all,
                             &b_10_ref, &af_bulk_zero, 0.67049555724866);
        bulk_test_range_fraction(&model_10, &s_dummy_obs, &af_single_all,
                             &b_10_ref, &af_bulk_zero, 0.67049555724866);
        bulk_test_ranges_sum(&model_5, &s_dummy_obs, &af_single_all,
                             &b_10_ref, vec![&af_bulk_not_one, &af_bulk_one]);
    }

    #[test]
    fn test_scbm_sc_het_germ() {

        let model = SingleCellBulkModel::new(2,1,100);

        // single cell is het against het germline
        let af_single = vec![AlleleFreq(0.5)];
        let af_bulk = ContinuousAlleleFreqs::left_exclusive( 0.25..0.75 );

        let single_obs = create_obs_vector(3, 3);
        let bulk_obs = create_obs_vector(3, 3);

        let pileup = create_test_snv_pileup(&model, &single_obs, &bulk_obs);

        println!("SCBM het: pileup.joint_prob(af_single, af_bulk): {}", pileup.joint_prob(&af_single, &af_bulk).exp() );
        assert_relative_eq!( pileup.joint_prob(&af_single, &af_bulk).exp(), 0.00019473983947767667, epsilon = 0.000000000000000001 );
        println!("SCBM het: pileup.marginal_prob: {}", pileup.marginal_prob().exp() );
        assert_relative_eq!( pileup.marginal_prob().exp(), 0.00027403381880824275, epsilon = 0.000000000000000001 );
        println!("SCBM het: pileup.posterior_prob: {}", pileup.posterior_prob(&af_single, &af_bulk).exp() );
        assert_relative_eq!( pileup.posterior_prob(&af_single, &af_bulk).exp(), 0.710641629287177, epsilon = 0.000000000000001 );
        let (sc, blk) = pileup.map_allele_freqs();
        println!("SCBM het: pileup.map_allele_freqs: ({} {})", sc, blk );
        assert_eq!( pileup.map_allele_freqs(), ( AlleleFreq(0.5), AlleleFreq(0.5) ) );

        // test maximum a posteriori allele frequency estimates with uneven allele observations
        let single_obs = create_obs_vector(3, 7);
        let bulk_obs = create_obs_vector(3, 7);

        let pileup = create_test_snv_pileup(&model, &single_obs, &bulk_obs);
        let (sc, blk) = pileup.map_allele_freqs();
        println!("SCBM het: pileup.map_allele_freqs: ({} {})", sc, blk );
        assert_eq!( pileup.map_allele_freqs(), ( AlleleFreq(0.5), AlleleFreq(0.7) ) );
    }

    #[test]
    fn test_scbm_sc_hom_ref_germ() {

        let model = SingleCellBulkModel::new(2,1,100);

        // single cell is hom ref against hom ref germline
        let af_single = vec![AlleleFreq(0.0)];
        let af_bulk = ContinuousAlleleFreqs::inclusive( 0.0..0.25 );

        let single_obs = create_obs_vector(5, 0);
        let bulk_obs = create_obs_vector(5, 0);

        let pileup = create_test_snv_pileup(&model, &single_obs, &bulk_obs);

        println!("SCBM hom ref: pileup.joint_prob(af_single, af_bulk): {}", pileup.joint_prob(&af_single, &af_bulk).exp() );
        assert_relative_eq!( pileup.joint_prob(&af_single, &af_bulk).exp(), 1.2409982197541034, epsilon = 0.0000000000000001 );
        println!("SCBM hom ref: pileup.marginal_prob: {}", pileup.marginal_prob().exp() );
        assert_relative_eq!( pileup.marginal_prob().exp(), 1.719859092473503, epsilon = 0.000000000000001 );
        println!("SCBM hom ref: pileup.posterior_prob: {}", pileup.posterior_prob(&af_single, &af_bulk).exp() );
        assert_relative_eq!( pileup.posterior_prob(&af_single, &af_bulk).exp(), 0.7215697060212639, epsilon = 0.0000000000000001 );
        let (sc, blk) = pileup.map_allele_freqs();
        println!("SCBM hom ref: pileup.map_allele_freqs: ({} {})", sc, blk );
        assert_eq!( pileup.map_allele_freqs(), ( AlleleFreq(0.0), AlleleFreq(0.0) ) );

        // test maximum a posteriori allele frequency estimates with alt reads
        let single_obs = create_obs_vector(57, 3);
        let bulk_obs = create_obs_vector(27, 3);

        let pileup = create_test_snv_pileup(&model, &single_obs, &bulk_obs);
        let (sc, blk) = pileup.map_allele_freqs();
        println!("SCBM hom ref: pileup.map_allele_freqs: ({} {})", sc, blk );
        assert_eq!( pileup.map_allele_freqs(), ( AlleleFreq(0.0), AlleleFreq(0.1) ) );
    }

    #[test]
    fn test_scbm_sc_hom_alt_germ() {

        let model = SingleCellBulkModel::new(2,1,100);

        // single cell is hom alt against hom alt germline
        let af_single = vec![AlleleFreq(1.0)];
        let af_bulk = ContinuousAlleleFreqs::left_exclusive( 0.75..1.0 );

        let single_obs = create_obs_vector(0, 5);
        let bulk_obs = create_obs_vector(0, 5);

        let pileup = create_test_snv_pileup(&model, &single_obs, &bulk_obs);

        println!("SCBM hom alt: pileup.joint_prob(af_single, af_bulk): {}", pileup.joint_prob(&af_single, &af_bulk).exp() );
        assert_relative_eq!( pileup.joint_prob(&af_single, &af_bulk).exp(), 1.2409982197541034, epsilon = 0.0000000000000001 );
        println!("SCBM hom alt: pileup.marginal_prob: {}", pileup.marginal_prob().exp() );
        assert_relative_eq!( pileup.marginal_prob().exp(), 1.7198590924735035, epsilon = 0.0000000000000001 );
        println!("SCBM hom alt: pileup.posterior_prob: {}", pileup.posterior_prob(&af_single, &af_bulk).exp() );
        assert_relative_eq!( pileup.posterior_prob(&af_single, &af_bulk).exp(), 0.7215697060212638, epsilon = 0.0000000000000001 );
        let (sc, blk) = pileup.map_allele_freqs();
        println!("SCBM hom alt: pileup.map_allele_freqs: ({} {})", sc, blk );
        assert_eq!( pileup.map_allele_freqs(), ( AlleleFreq(1.0), AlleleFreq(1.0) ) );

        // test maximum a posteriori allele frequency estimates with alt reads
        let single_obs = create_obs_vector(3, 57);
        let bulk_obs = create_obs_vector(3, 27);

        let pileup = create_test_snv_pileup(&model, &single_obs, &bulk_obs);
        let (sc, blk) = pileup.map_allele_freqs();
        println!("SCBM hom alt: pileup.map_allele_freqs: ({} {})", sc, blk );
        assert_eq!( pileup.map_allele_freqs(), ( AlleleFreq(1.0), AlleleFreq(0.9) ) );
    }

    #[test]
    fn test_scbm_sc_hom_alt_som() {

        let model = SingleCellBulkModel::new(2,1,100);

        // single cell is hom alt against het germline
        let af_single = vec![AlleleFreq(1.0)];
        let af_bulk = ContinuousAlleleFreqs::left_exclusive( 0.25..0.75 );

        let single_obs = create_obs_vector(0, 4);
        let bulk_obs = create_obs_vector(3, 1);

        let pileup = create_test_snv_pileup(&model, &single_obs, &bulk_obs);

        println!("SCBM hom alt: pileup.joint_prob(af_single, af_bulk): {}", pileup.joint_prob(&af_single, &af_bulk).exp() );
        assert_relative_eq!( pileup.joint_prob(&af_single, &af_bulk).exp(), 0.06996173990713889, epsilon = 0.0000000000000001 );
        println!("SCBM hom alt: pileup.marginal_prob: {}", pileup.marginal_prob().exp() );
        assert_relative_eq!( pileup.marginal_prob().exp(), 0.2267813569619291, epsilon = 0.0000000000000001 );
        println!("SCBM hom alt: pileup.posterior_prob: {}", pileup.posterior_prob(&af_single, &af_bulk).exp() );
        assert_relative_eq!( pileup.posterior_prob(&af_single, &af_bulk).exp(), 0.30849863870813554, epsilon = 0.00000000000000001 );
        let (sc, blk) = pileup.map_allele_freqs();
        println!("SCBM hom alt: pileup.map_allele_freqs: ({} {})", sc, blk );
        assert_eq!( pileup.map_allele_freqs(), ( AlleleFreq(1.0), AlleleFreq(0.25) ) );

        // test maximum a posteriori allele frequency estimates with alt reads
        let single_obs = create_obs_vector(4, 56);
        let bulk_obs = create_obs_vector(21, 9);

        let pileup = create_test_snv_pileup(&model, &single_obs, &bulk_obs);
        let (sc, blk) = pileup.map_allele_freqs();
        println!("SCBM hom alt: pileup.map_allele_freqs: ({} {})", sc, blk );
        assert_eq!( pileup.map_allele_freqs(), ( AlleleFreq(1.0), AlleleFreq(0.3) ) );
    }
}