twitcher 0.6.8

Find template switch mutations in genomic data
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
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
use assert_cmd::cargo::cargo_bin_cmd;
use bstr::{BStr, ByteSlice};

use crate::common::{twitcher_cmd, twitcher_with_ref};

mod common;

fn test_has_ts(file: &str, padding: u32) {
    let cmd = twitcher_with_ref(&format!(
        "tests/data/{file} --padding {padding} --cluster-min-records 1 -vvv --costs tests/old_test_costs.tsa"
    ));
    assert!(cmd.status.success());
    let stdout_str = cmd.stdout.as_bstr();
    let stderr_str = cmd.stderr.as_bstr();
    eprintln!("{stderr_str}");
    let records: Vec<_> = stdout_str
        .lines()
        .filter(|l| !l.starts_with(b"#"))
        .collect();
    for r in &records {
        println!("{}", r.as_bstr());
    }
    let has_ts = records.iter().any(|r| r.contains_str(b"CIGARETS"));
    assert!(has_ts);
}

#[test]
fn test_single_mnv() {
    test_has_ts("single_mnv.vcf", 10);
}

#[test]
fn test_single_insertion() {
    test_has_ts("single_insertion.vcf", 20);
}

#[test]
fn test_output_sorted() {
    let cmd = twitcher_with_ref("tests/data/sorted_output.vcf.gz --padding 20");
    assert!(cmd.status.success());
    let mut last_chr = None;
    let mut last_pos = 0;
    for record in cmd
        .stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#"))
    {
        let mut tokens = record.as_bstr().split_str(b"\t");
        let chr = tokens.next().unwrap();
        assert!(
            last_chr.is_none_or(|last| last == chr),
            "the chr is changing for an example where it shouldn't",
        );
        last_chr = Some(chr);
        let pos = tokens.next().unwrap().to_str().unwrap().parse().unwrap();
        assert!(pos >= last_pos, "The output is not sorted");
        last_pos = pos;
    }
}

fn test_region_output(filename: &str, expected: &str) {
    let dir = tempfile::tempdir().unwrap();
    let regions = format!("{}/{filename}", dir.path().to_str().unwrap());
    let mut cmd = cargo_bin_cmd!("twitcher");
    cmd.args([
        "vcf",
        "./test_files/test.vcf",
        "--reference",
        "./test_files/test.fa",
        "--output-regions",
        &regions,
        "--cluster-min-records",
        "1",
        "--costs",
        "./tests/old_test_costs.tsa",
    ]);
    let res = cmd.unwrap();
    let stderr_str = res.stderr.as_bstr();
    eprintln!("{stderr_str}");
    assert!(res.status.success());
    let output = std::fs::read_to_string(regions).unwrap();
    assert_eq!(output, expected);
}

#[test]
fn test_region_output_tab() {
    test_region_output("regions.some.ext", "chrZ\t101\t108\nchrZ\t201\t207\n");
}

#[test]
fn test_region_output_bed() {
    test_region_output("regions.bed", "chrZ\t100\t108\nchrZ\t200\t207\n");
}

// Tests below use ts_chr1_4304586.vcf — a real DRAGEN WGS cluster at chr1:4304586
// (a tandem-repeat deletion) that produces a reverse template switch with current default
// costs. Padding is 30 (the TS jump is -22 bp); this fits within the debug binary's memory.

const TS_FIXTURE: &str = "tests/data/ts_chr1_4304586.vcf --cluster-min-records 1 --padding 30";

#[test]
fn test_ts_detection_default_costs() {
    let cmd = twitcher_with_ref(TS_FIXTURE);
    let stderr_str = cmd.stderr.as_bstr();
    eprintln!("{stderr_str}");
    assert!(cmd.status.success());
    let has_ts = cmd
        .stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#"))
        .any(|r| r.contains_str(b"CIGARETS"));
    assert!(has_ts, "expected CIGARETS in output with default costs");
}

#[test]
fn test_no_ts_flag_suppresses_template_switches() {
    let cmd = twitcher_with_ref(&format!("{TS_FIXTURE} --no-ts"));
    let stderr_str = cmd.stderr.as_bstr();
    eprintln!("{stderr_str}");
    assert!(cmd.status.success());
    let has_ts = cmd
        .stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#"))
        .any(|r| r.contains_str(b"CIGARETS"));
    assert!(!has_ts, "expected no CIGARETS when --no-ts is set");
}

#[test]
fn test_only_realigned_outputs_ts_records_only() {
    let cmd = twitcher_with_ref(&format!("{TS_FIXTURE} --only-realigned"));
    let stderr_str = cmd.stderr.as_bstr();
    eprintln!("{stderr_str}");
    assert!(cmd.status.success());
    let data_lines: Vec<_> = cmd
        .stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#"))
        .collect();
    assert!(
        !data_lines.is_empty(),
        "expected at least one output record"
    );
    for line in &data_lines {
        assert!(
            line.contains_str(b"CIGARETS"),
            "non-CIGARETS record emitted with --only-realigned: {}",
            line.as_bstr()
        );
    }
}

#[test]
fn test_csv_output_created() {
    let dir = tempfile::tempdir().unwrap();
    let csv_path = dir.path().join("output.csv");
    let cmd = twitcher_with_ref(&format!(
        "{TS_FIXTURE} --csv {}",
        csv_path.to_str().unwrap()
    ));
    let stderr_str = cmd.stderr.as_bstr();
    eprintln!("{stderr_str}");
    assert!(cmd.status.success());
    assert!(csv_path.exists(), "CSV file was not created");
    let content = std::fs::read_to_string(&csv_path).unwrap();
    let mut lines = content.lines();
    let header = lines.next().expect("CSV must have a header row");
    assert!(!header.is_empty(), "CSV header must not be empty");
    assert!(
        lines.next().is_some(),
        "CSV must have at least one data row for the detected TS"
    );
}

#[test]
fn test_fpa_detects_template_switches() {
    // --fpa uses the Four-Point-Aligner instead of the default aligner.
    // It is faster but has limited capabilities; it must still find CIGARETS here.
    let cmd = twitcher_with_ref(&format!("{TS_FIXTURE} --fpa"));
    let stderr_str = cmd.stderr.as_bstr();
    eprintln!("{stderr_str}");
    assert!(cmd.status.success());
    let has_ts = cmd
        .stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#"))
        .any(|r| r.contains_str(b"CIGARETS"));
    assert!(has_ts, "expected CIGARETS in output with --fpa");
}

#[test]
fn test_no_panic_on_null_leading_match() {
    // Regression: leading_matches was unwrap()ed and could be None, causing a panic.
    // crash_chr4_leading_match.vcf.gz is the reproducer from chr4 that triggered it.
    // https://version.helsinki.fi/kraujasp/twitcher/-/work_items/58
    let cmd = twitcher_cmd(&[
        "vcf",
        "tests/data/crash_chr4_leading_match.vcf.gz",
        "--reference",
        "tests/data/hg38.chr4_part.fa.gz",
        "--threads",
        "1",
        "-vvv",
    ]);
    assert!(cmd.status.success());
}

#[test]
fn test_an_ac_correct_in_realigned_records() {
    // Regression for #52: AN was computed as `allele_count() - 1` (number of ALT allele
    // types) instead of total called alleles from genotypes. For a 0/1 call this gave
    // AN=1; for 1/1 it gave AN=1 with AC=2, triggering a bcftools validation error.
    let cmd = twitcher_with_ref(TS_FIXTURE);
    assert!(cmd.status.success());
    let realigned: Vec<_> = cmd
        .stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#") && l.contains_str(b"CIGARETS"))
        .collect();
    assert!(
        !realigned.is_empty(),
        "expected at least one realigned record"
    );
    for record in &realigned {
        let info = record
            .split_str(b"\t")
            .nth(7)
            .expect("VCF record must have INFO field");
        assert!(
            info.contains_str(b"AN=2"),
            "AN must be 2 (two called alleles from 0/1 genotype), got: {}",
            info.as_bstr()
        );
        assert!(
            info.contains_str(b"AC=1"),
            "AC must be 1 (one alt allele from 0/1 genotype), got: {}",
            info.as_bstr()
        );
    }
}

#[test]
fn test_phased_split_cluster_correlates_clustergrp() {
    // Two records on opposite haplotypes (0|1 + 1|0, same PS) form one proximity cluster.
    // After haplotype splitting: two 1-record sub-clusters, each independently realigned.
    // Both should produce a TS record sharing the same CLUSTERGRP value.
    let mut cmd = cargo_bin_cmd!("twitcher");
    cmd.args([
        "vcf",
        "./test_files/phased_split_cluster.vcf",
        "--reference",
        "./test_files/test.fa",
        "--cluster-min-records",
        "1",
        "--costs",
        "./tests/old_test_costs.tsa",
    ]);
    let res = cmd.unwrap();
    let stderr_str = res.stderr.as_bstr();
    eprintln!("{stderr_str}");
    assert!(res.status.success());

    let stdout = res.stdout.as_bstr();
    let ts_records: Vec<_> = stdout
        .lines()
        .filter(|l| !l.starts_with(b"#") && l.contains_str(b"CIGARETS"))
        .collect();

    assert_eq!(
        ts_records.len(),
        2,
        "expected two TS records, one per haplotype sub-cluster"
    );

    let clustergrps: Vec<_> = ts_records
        .iter()
        .filter_map(|rec| {
            rec.split_str(b"\t")
                .nth(7) // INFO field
                .and_then(|info| {
                    info.split_str(b";")
                        .find(|f| f.starts_with(b"CLUSTERGRP="))
                        .map(|f| f[b"CLUSTERGRP=".len()..].to_vec())
                })
        })
        .collect();
    assert_eq!(clustergrps.len(), 2, "both TS records must have CLUSTERGRP");
    assert_eq!(
        clustergrps[0], clustergrps[1],
        "TS records from the same proximity cluster must share CLUSTERGRP"
    );

    // Each sub-cluster must have received the correct per-haplotype GT.
    let gts: Vec<_> = ts_records
        .iter()
        .filter_map(|rec| {
            rec.split_str(b"\t")
                .nth(9) // sample column
                .and_then(|sample| sample.split_str(b":").next().map(<[u8]>::to_vec))
        })
        .collect();
    assert!(
        gts.iter().any(|g| g == b"0|1"),
        "hap=1 sub-cluster should produce 0|1 GT"
    );
    assert!(
        gts.iter().any(|g| g == b"1|0"),
        "hap=0 sub-cluster should produce 1|0 GT"
    );
}

#[test]
fn test_multiallelic_1_2_uses_per_haplotype_allele() {
    // A single phased multi-allelic 1|2 record splits into two haplotype sub-clusters:
    // H0 carries ALT allele 1, H1 carries ALT allele 2. Each must be realigned using its
    // own allele's sequence (not allele 1 for both) and emit a biallelic, correctly
    // oriented genotype. allele 1 = 7×T, allele 2 = 6×T so the two are distinguishable.
    let mut cmd = cargo_bin_cmd!("twitcher");
    cmd.args([
        "vcf",
        "./test_files/multiallelic_1_2_both_realign.vcf",
        "--reference",
        "./test_files/test.fa",
        "--cluster-min-records",
        "1",
        "--padding",
        "20",
        "--costs",
        "./tests/old_test_costs.tsa",
    ]);
    let res = cmd.unwrap();
    eprintln!("{}", res.stderr.as_bstr());
    assert!(res.status.success());

    let ts_records: Vec<_> = res
        .stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#") && l.contains_str(b"CIGARETS"))
        .collect();
    assert_eq!(ts_records.len(), 2, "expected one TS record per haplotype");

    // Collect (ALT, GT) per record.
    let pairs: Vec<(Vec<u8>, Vec<u8>)> = ts_records
        .iter()
        .map(|rec| {
            let cols: Vec<_> = rec.split_str(b"\t").collect();
            let alt = cols[4].to_vec();
            let gt = cols[9].split_str(b":").next().unwrap().to_vec();
            (alt, gt)
        })
        .collect();

    // H0: allele 1 (7×T) on the first slot -> 1|0.
    assert!(
        pairs
            .iter()
            .any(|(alt, gt)| alt == b"TTTTTTT" && gt == b"1|0"),
        "H0 must realign allele 1 (TTTTTTT) and emit 1|0, got {pairs:?}"
    );
    // H1: allele 2 (6×T) on the second slot -> 0|1. Proves allele 2's sequence was used.
    assert!(
        pairs
            .iter()
            .any(|(alt, gt)| alt == b"TTTTTT" && gt == b"0|1"),
        "H1 must realign allele 2 (TTTTTT) and emit 0|1, got {pairs:?}"
    );
}

/// Run the vcf subcommand on a `test_files/` fixture and return the non-header data lines.
fn run_vcf_data_lines(fixture: &str, extra: &[&str]) -> Vec<Vec<u8>> {
    let mut cmd = cargo_bin_cmd!("twitcher");
    cmd.args([
        "vcf",
        &format!("./test_files/{fixture}"),
        "--reference",
        "./test_files/test.fa",
        "--cluster-min-records",
        "1",
        "--padding",
        "20",
        "--costs",
        "./tests/old_test_costs.tsa",
    ]);
    cmd.args(extra);
    let res = cmd.unwrap();
    eprintln!("{}", res.stderr.as_bstr());
    assert!(!res.stderr.as_bstr().contains_str(b"oom"));
    assert!(res.status.success());
    res.stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#"))
        .map(<[u8]>::to_vec)
        .collect()
}

fn alt_and_gt(line: &[u8]) -> (Vec<u8>, Vec<u8>) {
    let cols: Vec<_> = line.split_str(b"\t").collect();
    let alt = cols[4].to_vec();
    let gt = cols[9].split_str(b":").next().unwrap().to_vec();
    (alt, gt)
}

#[test]
fn test_partial_realign_reconstructs_other_haplotype() {
    // 1|2 with allele 1 = TTTTTTT (realigns to a TS), allele 2 = CAAAAAC (no TS). The H1
    // sub-cluster must be reconstructed as a biallelic record (REF + allele 2, 0|1), NOT
    // re-emitted as the raw multi-allelic 1|2 line, and there must be no duplicate.
    let lines = run_vcf_data_lines("multiallelic_1_2.vcf", &[]);
    assert_eq!(lines.len(), 2, "one TS + one reconstructed record");
    let ts: Vec<_> = lines
        .iter()
        .filter(|l| l.contains_str(b"CIGARETS"))
        .collect();
    assert_eq!(ts.len(), 1, "exactly one TS record");
    let pairs: Vec<_> = lines.iter().map(|l| alt_and_gt(l)).collect();
    assert!(pairs.iter().any(|(a, g)| a == b"TTTTTTT" && g == b"1|0"));
    assert!(pairs.iter().any(|(a, g)| a == b"CAAAAAC" && g == b"0|1"));
    // No multi-allelic (comma) ALT and no duplicate of the raw original survived.
    assert!(
        lines
            .iter()
            .all(|l| !l.split_str(b"\t").nth(4).unwrap().contains_str(b","))
    );
}

#[test]
fn test_no_ts_split_cluster_emits_no_duplicates() {
    // 1|2 where neither allele realigns. Both haplotypes are reconstructed as biallelic
    // records (allele 1 -> 1|0, allele 2 -> 0|1); the raw original must not appear, and
    // nothing is emitted twice.
    let lines = run_vcf_data_lines("multiallelic_1_2_no_ts.vcf", &[]);
    assert_eq!(lines.len(), 2, "two reconstructed records, no duplicates");
    assert!(lines.iter().all(|l| !l.contains_str(b"CIGARETS")));
    let pairs: Vec<_> = lines.iter().map(|l| alt_and_gt(l)).collect();
    assert!(pairs.iter().any(|(a, g)| a == b"C" && g == b"1|0"));
    assert!(pairs.iter().any(|(a, g)| a == b"G" && g == b"0|1"));
}

#[test]
fn test_hom_alt_non_one_allele_emitted_once() {
    // A 2/2 record is homozygous for ALT allele 2. It must be emitted exactly once as a
    // biallelic 1/1 record using allele 2's sequence, not split across haplotypes.
    let lines = run_vcf_data_lines("hom_alt_2_2.vcf", &[]);
    assert_eq!(lines.len(), 1, "hom-alt emitted exactly once");
    let (alt, gt) = alt_and_gt(&lines[0]);
    assert_eq!(alt, b"G", "must use allele 2's sequence");
    assert_eq!(gt, b"1/1");
}

#[test]
fn test_twitcher_phase_flag_marks_changed_genotypes() {
    // Splitting 1|2 into two biallelic records rewrites the GT of both, so both must carry
    // the TWITCHERPHASE flag.
    let changed = run_vcf_data_lines("multiallelic_1_2_no_ts.vcf", &[]);
    assert_eq!(changed.len(), 2);
    assert!(
        changed.iter().all(|l| l.contains_str(b"TWITCHERPHASE")),
        "records with a rewritten GT must be flagged, got {:?}",
        changed.iter().map(BStr::new).collect::<Vec<_>>()
    );

    // Two already-phased records that pass through with their GT/PS intact must not be
    // flagged.
    let unchanged = run_vcf_data_lines("phased_split_cluster.vcf", &[]);
    assert!(
        unchanged.iter().all(|l| !l.contains_str(b"TWITCHERPHASE")),
        "records with unchanged GT/PS must not be flagged, got {:?}",
        unchanged.iter().map(BStr::new).collect::<Vec<_>>()
    );
}

#[test]
fn test_phased_split_preserves_slot_to_haplotype_orientation() {
    // 2|1, PS=42: allele 1 = TTTTTTT (7T), allele 2 = TTTTTT (6T). VCF phasing maps GT slot
    // 0 -> haplotype 0, slot 1 -> haplotype 1, consistently across the whole phaseset. So the
    // split must keep slot 0's allele (allele 2, 6T) on the 1|0 record and slot 1's allele
    // (allele 1, 7T) on the 0|1 record — driven by GT slot, NOT numeric allele order — and
    // carry PS=42 on both, so the records stay phase-consistent with any other PS=42 record
    // far outside this cluster.
    let lines = run_vcf_data_lines("phased_2_1_orientation.vcf", &[]);
    assert_eq!(lines.len(), 2);

    // (ALT, GT, PS) per record.
    let recs: Vec<(Vec<u8>, Vec<u8>, Vec<u8>)> = lines
        .iter()
        .map(|l| {
            let cols: Vec<_> = l.split_str(b"\t").collect();
            let mut sample = cols[9].split_str(b":");
            let gt = sample.next().unwrap().to_vec();
            let ps = sample.next().unwrap().to_vec();
            (cols[4].to_vec(), gt, ps)
        })
        .collect();

    // slot 0 (haplotype 0) carries allele 2 (6T) -> 1|0
    assert!(
        recs.iter()
            .any(|(alt, gt, ps)| alt == b"TTTTTT" && gt == b"1|0" && ps == b"42"),
        "1|0 must carry slot-0 allele (allele 2, TTTTTT) with PS=42, got {recs:?}"
    );
    // slot 1 (haplotype 1) carries allele 1 (7T) -> 0|1
    assert!(
        recs.iter()
            .any(|(alt, gt, ps)| alt == b"TTTTTTT" && gt == b"0|1" && ps == b"42"),
        "0|1 must carry slot-1 allele (allele 1, TTTTTTT) with PS=42, got {recs:?}"
    );
}

/// Run the vcf subcommand on a `test_files/` fixture with default clustering settings
/// (no `--cluster-min-records` override) and return the non-header data lines.
fn run_vcf_default_data_lines(fixture: &str) -> Vec<Vec<u8>> {
    let mut cmd = cargo_bin_cmd!("twitcher");
    cmd.args([
        "vcf",
        &format!("./test_files/{fixture}"),
        "--reference",
        "./test_files/test.fa",
        "--costs",
        "./tests/old_test_costs.tsa",
    ]);
    let res = cmd.unwrap();
    eprintln!("{}", res.stderr.as_bstr());
    assert!(res.status.success());
    res.stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#"))
        .map(<[u8]>::to_vec)
        .collect()
}

#[test]
fn test_no_alt_genotypes_ignored_for_clustering() {
    // A single 0|1 het accompanied by no-alt genotypes (0|0, .|., 0|., .|0) must produce the
    // exact same output as the het alone: those genotypes carry no allele on any haplotype, so
    // they must not be considered for clustering. Regression for a case where a nearby no-alt
    // record turned a lone variant into a proximity cluster, routed it into haplotype splitting,
    // and the resulting lone sub-cluster was then silently dropped — losing the real variant.
    let with = run_vcf_default_data_lines("noalt_genotypes_with.vcf");
    let without = run_vcf_default_data_lines("noalt_genotypes_without.vcf");
    assert_eq!(
        with, without,
        "no-alt genotypes must not change the output vs. running without them"
    );
    assert_eq!(without.len(), 1, "the lone 0|1 het must be preserved");
    assert!(without[0].contains_str(b"0|1"));
}

#[test]
fn test_phasing_bam_flag_accepted_and_pipeline_succeeds() {
    // Smoke test: --phasing-bam is wired through the full pipeline.
    // The reads BAM has different chromosomes from the test VCF, so the fetch
    // returns no reads and phasing is a no-op; the output must match the
    // no-BAM output record count.
    let without_bam = twitcher_with_ref(TS_FIXTURE);
    let with_bam = twitcher_with_ref(&format!(
        "{TS_FIXTURE} --phasing-bam tests/data/reads/test.bam"
    ));
    let stderr = with_bam.stderr.as_bstr();
    eprintln!("{stderr}");
    assert!(with_bam.status.success(), "--phasing-bam must not crash");
    let count_without = without_bam
        .stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#"))
        .count();
    let count_with = with_bam
        .stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#"))
        .count();
    assert_eq!(
        count_without, count_with,
        "--phasing-bam with no overlapping reads must not change record count"
    );
}

#[test]
fn test_target_region_empty_range_produces_no_records() {
    // chr1:1-10 contains no variants in the fixture; output should be header-only.
    let cmd = twitcher_with_ref(&format!("{TS_FIXTURE} -t chr1:1-10"));
    let stderr_str = cmd.stderr.as_bstr();
    eprintln!("{stderr_str}");
    assert!(cmd.status.success());
    let data_count = cmd
        .stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#"))
        .count();
    assert_eq!(
        data_count, 0,
        "target region chr1:1-10 should yield no records"
    );
}

/// Run the vcf subcommand on a `test_files/` fixture, keeping stderr so tests can assert on
/// the diagnostics of the run.
fn run_vcf_fixture(fixture: &str) -> std::process::Output {
    let mut cmd = cargo_bin_cmd!("twitcher");
    cmd.args([
        "vcf",
        &format!("./test_files/{fixture}"),
        "--reference",
        "./test_files/test.fa",
        "--cluster-min-records",
        "1",
        "--padding",
        "20",
        "--costs",
        "./tests/old_test_costs.tsa",
    ]);
    let res = cmd.unwrap();
    eprintln!("{}", res.stderr.as_bstr());
    assert!(res.status.success());
    res
}

#[test]
fn test_star_allele_covered_by_deletion_in_same_cluster() {
    // `*` marks an allele that is missing because an upstream deletion spans the position, so
    // it describes no sequence change of its own. Here that deletion (101 AAAAAA -> A) sits on
    // the same haplotype and in the same cluster, so the alt sequence must be built from the
    // deletion alone. Before, the `*` record tripped the overlap check — it overlaps the
    // deletion by construction — and the whole cluster was skipped.
    let res = run_vcf_fixture("star_allele_upstream_deletion.vcf");
    let stderr = res.stderr.as_bstr();
    assert!(
        !stderr.contains_str(b"Overlapping mutation"),
        "the `*` record must not count as an overlapping mutation"
    );
    let lines: Vec<_> = res
        .stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#"))
        .collect();
    assert_eq!(lines.len(), 2, "both variants must survive");
    let pairs: Vec<_> = lines.iter().map(|l| alt_and_gt(l)).collect();
    // The deletion stays on haplotype 0, the SNV of the `*` record's other allele on
    // haplotype 1. The `*` itself is fully accounted for by the deletion and disappears.
    assert!(
        pairs.iter().any(|(a, g)| a == b"A" && g == b"1|0"),
        "deletion must be kept on haplotype 0, got {pairs:?}"
    );
    assert!(
        pairs.iter().any(|(a, g)| a == b"T" && g == b"0|1"),
        "the non-`*` allele must be kept on haplotype 1, got {pairs:?}"
    );
}

#[test]
fn test_star_allele_without_spanning_deletion_in_cluster() {
    // The spanning deletion is not part of this cluster (it lies outside the scanned window),
    // so the `*` haplotype has no other record to explain it. The `*` must still not be
    // spliced into the alt sequence — that used to fail the alignment with
    // "query sequence contains an unsupported ASCII character: *" — while the record itself
    // is passed through unchanged rather than dropped.
    let res = run_vcf_fixture("star_allele_alone.vcf");
    let stderr = res.stderr.as_bstr();
    assert!(
        !stderr.contains_str(b"unsupported ASCII character"),
        "`*` must never reach the aligner"
    );
    let lines: Vec<_> = res
        .stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#"))
        .collect();
    let pairs: Vec<_> = lines.iter().map(|l| alt_and_gt(l)).collect();
    assert!(
        pairs.iter().any(|(a, g)| a == b"T" && g == b"1|0"),
        "the real allele must be kept, got {pairs:?}"
    );
    assert!(
        pairs.iter().any(|(a, g)| a == b"*" && g == b"0|1"),
        "the `*` allele must be passed through, not dropped, got {pairs:?}"
    );
}

#[test]
fn test_symbolic_allele_is_not_spliced_into_alt_sequence() {
    // Symbolic alleles carry no literal sequence. Splicing `<NON_REF>` in used to fail the
    // alignment on a non-DNA character; it must be skipped with a warning and the record
    // passed through unchanged.
    let res = run_vcf_fixture("symbolic_allele.vcf");
    let stderr = res.stderr.as_bstr();
    assert!(
        !stderr.contains_str(b"unsupported ASCII character"),
        "`<NON_REF>` must never reach the aligner"
    );
    assert!(
        stderr.contains_str(b"is not a DNA sequence"),
        "skipping a symbolic allele must be reported"
    );
    let lines: Vec<_> = res
        .stdout
        .as_bstr()
        .lines()
        .filter(|l| !l.starts_with(b"#"))
        .collect();
    assert_eq!(lines.len(), 1, "the record must be passed through");
    assert!(lines[0].contains_str(b"<NON_REF>"));
}