This file is indexed.

/usr/bin/maq.pl is in maq 0.7.1-5.

This file is owned by root:root, with mode 0o755.

The actual contents of the file can be viewed below.

  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
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
#!/usr/bin/perl -w

# Author: lh3

use strict;
use warnings;
use Getopt::Std;
use File::Copy;
use File::Spec;
use Cwd qw/getcwd abs_path/;

my $version = '0.4.7';
&usage if (@ARGV < 1);
# global variables
my @nucl_type = (0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4);
my @conv = split('', 'XACMGRSVTWYHKDBN');

my $command = shift(@ARGV);
my %func = (easyrun=>\&easyrun, chrpt2snp=>\&chrpt2snp, fastq2bfq=>\&fastq2bfq,
			cat2pair=>\&cat2pair, SNPfilter=>\&SNPfilter, ucsc2snp=>\&ucsc2snp,
			statmap=>\&statmap, sv=>\&sv, demo=>\&demo, splitread=>\&splitread,
			indelpe=>\&indelpe);

die("Unknown command \"$command\".\n") if (!defined($func{$command}));
&{$func{$command}}();
exit(0);

#
# easyrun command
#

sub easyrun
{
	my %opts = (d=>'easyrun', 1=>0, 2=>0, A=>'', n=>2000000, a=>250, e=>3, q=>40, m=>2,
				D=>256, E=>20, N=>2, w=>5, b=>60, B=>2);
	getopts('d:1:2:A:n:a:pe:q:E:D:m:N:b:B:S', \%opts);
	die qq(
Usage:   maq.pl easyrun [options] <ref.fa|bfa> <reads1.fq|bfq> [reads2.fastq]

Options: -d DIR    output directory [$opts{d}]
         -n INT    number of reads in a split file [$opts{n}]
         -1 INT    length of read1 [$opts{1}]

         -A FILE   file that contains the 3'adapter [null]
         -m INT    number of mismatches in the seed [$opts{m}]
         -N INT    number of haplotypes [$opts{N}]
         -b INT    maximum sum of qualities of mismatches [$opts{b}]

         -e INT    minimum read depth to call a SNP [$opts{e}]
         -D INT    maximum read depth to call a SNP [$opts{D}]
         -E INT    minimum neighbour consensus quality to call a SNP [$opts{E}]
         -q INT    quality threshold for the final SNP calls [$opts{q}]
         -w INT    SNP within +/- INT distant to an indel to be filtered [$opts{w}]
         -B INT    maximum number of SNPs in a window [$opts{B}]
         -S        apply indel detection with the splitread (VERY SLOW)

         -p        mate-pair alignment (PE mode)
         -2 INT    length of read2 (PE only) [$opts{2}]
         -a INT    max insert size (PE only) [$opts{a}]

Notes: For pooled data, please specify the number of haplotypes in the
       pool with `-N' and set `-E 0' at the same time. Do not try
       `-N 1', which only causes segfault.

) if (@ARGV < 2);
	if (defined($opts{p}) && @ARGV != 3) {
	  die("** for PE mode, two reads files should be specified.\n");
	}
	my $d = $opts{d};
	my $cwd = getcwd;
	my $mapopt = " -n $opts{m} -e ".($opts{b}+10);
	map { $_ = File::Spec->rel2abs($_) } @ARGV; # change to absolute path
    $mapopt .= $opts{1}? " -1 $opts{1}" : '';
    $mapopt .= $opts{2}? " -2 $opts{2}" : '';
    $mapopt .= $opts{A}? " -d " . File::Spec->rel2abs($opts{A}) : '';
	my $filtopt = "-d $opts{e} -D $opts{D} -n $opts{E}";
	my $exe = gwhich('maq') || die("** Cannot find 'maq' executable.");
	my $pl = gwhich($0) || die("** Cannot find 'maq.pl' script. Should be a bug actually.");
	mkdir($d) unless (-d $d);
	# run fasta2bfa
	my ($is_gzip, $first) = &test_file($ARGV[0]);
	if (ord($first) == ord('>')) { # fasta format
	  if ($is_gzip) { # gzipped
		&run_cmd("gzip -dc $ARGV[0] | $exe fasta2bfa - $d/ref.bfa 2> /dev/null");
	  } else { # plain
		&run_cmd("$exe fasta2bfa $ARGV[0] $d/ref.bfa 2> /dev/null");
	  }
	} else { # create a symbolic link
	  &run_cmd("ln -sf $ARGV[0] $d/ref.bfa");
	}
	# run fastq2bfq && match
	shift(@ARGV);
	my $i = 0;
	my (@map_files, $map_file);
	foreach my $file (@ARGV) {
	  ++$i;
	  ($is_gzip, $first) = &test_file($file);
	  if (ord($first) == ord('@')) { # fastq format
		if ($is_gzip) { # gzipped
		  &run_cmd("gzip -dc $file | $exe fastq2bfq -n $opts{n} - $d/read$i");
		} else { # plain fastq
		  &run_cmd("$exe fastq2bfq -n $opts{n} $file $d/read$i");
		}
	  } elsif ($is_gzip) { # probably bfq
		&run_cmd("ln -sf $file $d/read$i.bfq");
	  } else {
		die("** Cannot guess the format of file '$file'.");
	  }
	}
	# run alignment
	if (defined $opts{p}) { # paired reads
	  my @bfq_files = `(cd $d; find . -name "read1*.bfq")`;
	  foreach (@bfq_files) {
		chomp;
		/^\.\/read1(.*)\.bfq$/;
		my $tag = $1;
		my $bfqs = (-f "$d/read2$tag.bfq")? "read1$tag.bfq read2$tag.bfq" : "read1$tag.bfq";
		&run_cmd("(cd $d; $exe map $mapopt aln$tag.map ref.bfa $bfqs 2> aln$tag.map.log)");
		push(@map_files, "aln$tag.map");
	  }
	} else { # unpaired reads
	  my @bfq_files = `(cd $d; find . -name "read*.bfq")`;
	  foreach (@bfq_files) {
		chomp;
		/^\.\/read(.*)\.bfq$/;
		my $tag = $1;
		&run_cmd("(cd $d; $exe map $mapopt -u unmap$tag.txt aln$tag.map ref.bfa read$tag.bfq 2> aln$tag.map.log)");
		push(@map_files, "aln$tag.map");
	  }
	}
	if (@map_files > 1) { # we should merge them
	  &run_cmd("(cd $d; $exe mapmerge all.map @map_files)");
	} else {
	  &run_cmd("(cd $d; mv $map_files[0] all.map)");
	}
	$map_file = "all.map";
	&run_cmd("(cd $d; $exe mapcheck ref.bfa $map_file > mapcheck.txt)");
	&run_cmd("(cd $d; $exe assemble -N $opts{N} -Q $opts{b} consensus.cns ref.bfa $map_file 2> assemble.log)");
	&run_cmd("$exe cns2fq $d/consensus.cns > $d/cns.fq");
	&run_cmd("$exe cns2snp $d/consensus.cns > $d/cns.snp");
	&run_cmd("$exe cns2win $d/consensus.cns > $d/cns.win");
	&run_cmd("$exe indelsoa $d/ref.bfa $d/$map_file > $d/cns.indelse");
	if (defined $opts{p}) {
	  &run_cmd("$exe indelpe $d/ref.bfa $d/$map_file > $d/cns.indelpe");
	  &run_cmd("$pl SNPfilter -q $opts{q} -w $opts{w} -N $opts{B} -f $d/cns.indelse -F $d/cns.indelpe $filtopt -Q60 $d/cns.snp > $d/cns.final.snp");
	} else {
	  &run_cmd("(cd $d; touch unmap.indel)");
	  if (!defined($opts{S})) {
		&run_cmd("$pl SNPfilter -q $opts{q} -w $opts{w} -N $opts{B} -f $d/cns.indelse $filtopt $d/cns.snp > $d/cns.final.snp");
	  } else { # call splitread indels
		&run_cmd("(cd $d; cat unmap*.txt | $pl splitread -p unmap - ref.bfa 2> unmap.log)");
		&run_cmd("$pl SNPfilter -q $opts{q} -w $opts{w} -N $opts{B} -f $d/cns.indelse -S $d/unmap.indel $filtopt $d/cns.snp > $d/cns.final.snp");
	  }
	}
	&run_cmd("(cd $d; ln -s cns.final.snp cns.filter.snp)");
#	my $maq_plot = gwhich("maq_plot.pl");
#	my $gnuplot = gwhich("gnuplot");
#	if ($maq_plot && $gnuplot) {
#		&run_cmd("$maq_plot depth -x 0.2 $d/depth $d/cns.win");
#	}
	&run_cmd("$pl statmap $d/*.map.log");
}

sub statmap {
  my ($is_paired, $n_reads1, $n_reads2, $n_moved_high, $n_moved_low, $n_mapped1, $n_mapped2, $n_paired, $n_added);
  $is_paired = $n_reads1 = $n_reads2 = $n_moved_high = $n_moved_low = $n_mapped1 = $n_mapped2 = $n_paired = $n_added = 0;

  while (<>) {
	if (/^--\s*\(total,\s*isPE,.*\) .*=\s*\((\d+),\s*(\d+),\s*(\d+),\s*(\d+)\)/) {
	  $is_paired = $2;
	  if ($is_paired) { $n_reads2 += $1; $n_mapped2 += $3; }
	  else { $n_reads1 += $1; $n_mapped1 += $3; }
	  $n_paired += $4;
	} elsif (/match_data2mapping.* (\d+) pairs are added/) {
	  $n_added += $1;
	} elsif (/match_data2mapping.* first.*\((\d+), (\d+)\).*second.*\((\d+), (\d+)\)/) {
	  $n_moved_high += $1+$3;
	  $n_moved_low += $2+$4;
	}
  }
  my $tot = $n_reads1 + $n_reads2;
  my $map1_ratio = ($n_reads1 == 0)? 'NA' : int(10000 * $n_mapped1 / $n_reads1) / 100;
  my $map2_ratio = ($n_reads2 == 0)? 'NA' : int(10000 * $n_mapped2 / $n_reads2) / 100;
  my $pair_ratio = ($n_mapped2 == 0)? 'NA' : int(10000 * $n_paired / $n_mapped2) / 100;
  my $add_ratio = ($n_paired == 0)? 'NA' : int(10000 * $n_added / $n_paired) / 100;
  my $high_ratio = ($n_paired == 0)? 'NA' : int(10000 * $n_moved_high / $n_paired) / 100;
  my $low_ratio = ($n_paired == 0)? 'NA' : int(10000 * $n_moved_low / $n_paired) / 100;
  print qq(
-- == statmap report ==\n
-- # single end (SE) reads: $n_reads1
-- # mapped SE reads: $n_mapped1 (/ $n_reads1 = $map1_ratio%)
-- # paired end (PE) reads: $n_reads2
-- # mapped PE reads: $n_mapped2 (/ $n_reads2 = $map2_ratio%)
-- # reads that are mapped in pairs: $n_paired (/ $n_mapped2 = $pair_ratio%)
-- # Q>=30 reads that are moved to meet mate-pair requirement: $n_moved_high (/ $n_paired = $high_ratio%)
-- # Q<30 reads that are moved to meet mate-pair requirement: $n_moved_low ($low_ratio%)
\n);
}

#
# SNPfilter command
#

sub SNPfilter {
  my %opts = (f=>'', S=>'', s=>3, m=>1, Q=>40, d=>3, F=>'', w=>5, D=>256, N=>2, W=>10, n=>20, c=>-1.0, q=>20);
  getopts('af:s:m:Q:d:D:w:F:W:N:c:n:S:q:', \%opts);
  die(qq{
Usage:   maq.pl SNPfilter [options] <cns2snp.snp>

Options: -d INT        minimum depth to call a SNP [$opts{d}]
         -D INT        maximum depth (<=254), otherwise ignored [$opts{D}]
         -n INT        minimum neighbouring quality [$opts{n}]
         -Q INT        required max mapping quality of the reads covering the SNP [$opts{Q}]
         -q INT        minimum consensus quality [$opts{q}]
         -w INT        size of the window in which SNPs should be filtered out [$opts{w}]
         -S FILE       splitread output [null]
         -F FILE       indelpe output [null]
         -f FILE       indelsoa output [null]
         -s INT        indelsoa score (= left_clip + right_clip - across) [$opts{s}]
         -m INT        indelsoa: max number of reads mapped across the indel [$opts{m}]
         -W INT        window size for filtering dense SNPs [$opts{W}]
         -N INT        maximum number of SNPs in a window [$opts{N}]
         -a            alternative filter for single end reads
\n}) unless (@ARGV);
  my (%hash, $fh);
  my $skip = $opts{w};
  if ($opts{f}) { # for indelsoa
	my $n = 0;
	open($fh, $opts{f}) || die;
	while (<$fh>) {
	  my @t = split;
	  next unless ($t[4]+$t[5]-$t[3] >= $opts{s} && $t[3] <= $opts{m}); # a simple filter
	  ++$n;
	  if ($t[2] < 0) { # potential deletion
		for (my $x = $t[1] + $t[2] - $skip; $x <= $t[1] + $skip; ++$x) {
		  $hash{$t[0],$x} = 1;
		}
	  } else { # potential insertion
		for (my $x = $t[1] - $skip; $x <= $t[1] + $t[2] + $skip; ++$x) {
		  $hash{$t[0],$x} = 1;
		}
	  }
	}
	close($fh);
	warn("-- $n potential soa-indels pass the filter.\n");
  }
  if ($opts{F}) { # for indelpe
	my $n = 0;
	open($fh, $opts{F}) || die;
	while (<$fh>) {
	  my @t = split;
	  next unless ($t[2] eq '*' || $t[2] eq '+');
	  ++$n;
	  for (my $x = $t[1] - 1; $x < $t[1] + 2*$skip; ++$x) {
		$hash{$t[0],$x} = 1;
	  }
	}
	close($fh);
	warn("-- $n potential pe-indels pass the filter.\n");
  }
  if ($opts{S}) { # for splitread
	my $n = 0;
	open($fh, $opts{S}) || die;
	while (<$fh>) {
	  next unless (/^VR/);
	  my @t = split;
	  next if ($t[5] < 2 || ($t[5] < 4 && $t[4] == 0));
	  ++$n;
	  $hash{$t[1],$_} = 1 for ($t[2] .. $t[3]);
	}
	close($fh);
	warn("-- $n potential splitread-indels pass the filter.\n");
  }
  my $is_alter = defined($opts{a});
  my (@last, $last_chr);
  $last_chr = '';
  while (<>) {
	my @t = split;
	next if ($hash{$t[0],$t[1]});
	my $is_good;
	if (!$is_alter) { # the default filter
	  $is_good = ($t[5] >= $opts{d} && $t[5] <= $opts{D} && $t[6] > $opts{c} && $t[7] >= $opts{Q} && $t[8] >= $opts{n})? 1 : 0;
	} else { # the alternative filter for SE reads
	  $is_good = ($t[5] >= $opts{d} && $t[5] <= $opts{D} && $t[6] > $opts{c} && $t[6] <= 4.0 && $t[8] >= $opts{n})? 1 : 0;
	}
	$is_good = 0 unless ($t[4] >= $opts{q} || ($t[2] ne $t[9] && $t[4]+$t[10] >= $opts{q})); # consensus quality filter
	next unless ($is_good); # drop
	if ($t[0] ne $last_chr) { # a different chr, print
	  map { print $_->{L} if ($_->{F}) } @last;
	  @last = ();
	  $last_chr = $t[0];
	}
	# The following block provided by Nathans Weeks. Thanks, Nathans.
	push(@last, {L => $_, X => $t[1], F => 1}); # Enqueue current SNP
	if ($#last == $opts{N}) {                   # number of SNPs in queue is N+1
	  if ($last[$#last]{X} - $last[0]{X} < $opts{W}) { # if all within window W
		map {$_->{F} = 0} @last; # all SNPs in the window of size W are "bad"
	  }
	  print STDOUT $last[0]{L} if ($last[0]{F}); # print first SNP if good
	  shift @last # dequeue first SNP
	}
  }
  # print the last few lines if applicable
  map { print $_->{L} if ($_->{F}) } @last;
}

#
# splitread
#

sub splitread {
  my %opts = (1=>12, 2=>12, p=>'splitread', a=>20);
  getopts('1:2:p:a:', \%opts);
  die(qq/
Usage:   maq.pl splitread [options] <input.unmap> <in.bfa>

Options: -1 INT     length of read1 [$opts{1}]
         -2 INT     length of read2 [$opts{2}]
         -a INT     maximum indel size [$opts{a}]
         -p STR     prefix [$opts{p}]

/) if (@ARGV < 2);
  my $maq = gwhich('maq') || die("[splitread] fail to locate 'maq'. Abort!\n");
  my ($l1, $l2) = ($opts{1}, $opts{2});
  my ($n, $fh, $fh1, $fh2);
  # generate fastq
  warn("-- generate paired end reads\n");
  open($fh, ($ARGV[0]=~/\.gz$/)? "gzip -dc $ARGV[0]" : $ARGV[0]) || die;
  open($fh1, ">$opts{p}1.fq") || die;
  open($fh2, ">$opts{p}2.fq") || die;
  while (<$fh>) {
	my @t = split;
	next if ($t[1] != 99);
	my $l = length($t[2]);
	die("[splitread] reads are too short!\n") if ($l < $l1 + $l2);
	# sequence
	my ($s1, $s2) = (substr($t[2], 0, $l1), substr($t[2], -$l2));
	# discard poor reads
	next if ($s1 =~ /N/i || $s2 =~ /N/i);
	$n = ($s1 =~ tr/A/A/);
	next if ($n == $l1);
	$n = ($s2 =~ tr/A/A/);
	next if ($n == $l2);
	# reverse $s2
	$s2 = reverse($s2);
	$s2 =~ tr/ACGTN/TGCAN/;
	# quality
	my ($q1, $q2) = (substr($t[3], 0, $l1), substr($t[3], -$l2));
	$q2 = reverse($q2);
	# output
	print $fh1 "\@$t[0]_$l/1\n$s1\n+\n$q1\n";
	print $fh2 "\@$t[0]_$l/2\n$s2\n+\n$q2\n";
  }
  close($fh); close($fh1); close($fh2);
  # paired end alignment
  my $ins_size = $l1 + $l2 + $opts{a};
  &run_cmd("$maq fastq2bfq $opts{p}1.fq $opts{p}1.bfq");
  &run_cmd("$maq fastq2bfq $opts{p}2.fq $opts{p}2.bfq");
  &run_cmd("$maq map -Pn1 -a $ins_size $opts{p}.map $ARGV[1] $opts{p}1.bfq $opts{p}2.bfq");
  # get indel sites
  my @last = ('', 0, 0, 0, 0); # name, begin, end, 
  open($fh, "$maq mapview $opts{p}.map |") || die;
  open($fh1, ">$opts{p}.indel") || die;
  print $fh1 qq{CC\tCC comments
CC\tRD read_id chr begin end ins_size
CC\tVR chr begin end ins_size n_reads
CC
CC\tIn VR lines, zero ins_size indicates that there are indels of different
CC\tlengths. In addition, indels supported by one read is not reliable.
CC
};
  while (<$fh>) {
	my @t = split;
	next if ($t[5] != 18 || $t[8] == 0 || $t[4] < 0);
	$t[0] =~ /^(\S+)_(\d+)\/[12]$/;
	my ($name, $l) = ($1, $2);
	next if ($t[4] == $l); # no indel
	if ($last[0] ne $t[1]) {
	  if ($last[0]) {
		print $fh1 join("\t", "VR", @last), "\n";
	  }
	  @last = ($t[1], 0, 0, 0, 0);
	}
	my ($b, $e, $s) = ($t[2], $t[2]+$t[4]-1, $l - $t[4]);
	if ($last[1] == 0) { # the first
	  @last = ($t[1], $b, $e, $s, 1);
	} elsif ($b <= $last[2]) { # overlap
	  if ($s != $last[3]) { # conflict
		$last[3] = 0;
	  }
	  ++$last[4];
	  $last[1] = $b;
	} else { # non-overlap
	  print $fh1 join("\t", "VR", @last), "\n";
	  @last = ($t[1], $b, $e, $s, 1);
	}
	print $fh1 join("\t", 'RD', $name, $t[1], $b, $e, $s), "\n";
  }
  close($fh); close($fh1);
}

#
# for Sanger's PE read format only
#

sub cat2pair
{
  my %opts = (1=>0, 2=>0);
  getopts('1:2:', \%opts);
  die qq(Usage: maq.pl cat2pair [-1 $opts{1}] [-2 $opts{2}] <read1_len> <input.fastq> [output.fastq]\n) if (@ARGV < 2);
  my ($tl1, $tl2) = ($opts{1}, $opts{2});
  my $size1 = $ARGV[0];
  my $fn = $ARGV[1];
  my $fn_out = (@ARGV >= 3)? $ARGV[2] : $ARGV[1];
  $fn_out =~ s/^.*\/([^\/\s]+)$/$1/ if ($fn_out =~ /\//);
  $fn_out =~ s/\.gz$// if ($fn_out =~ /\.gz$/);
  mkdir("read1"); mkdir("read2");
  my ($fh1, $fh2, $fhin);
  $fn = "gzip -dc $fn |" if ($fn =~ /\.gz$/);
  open($fhin, $fn) || die;
  open($fh1, ">read1/$fn_out") || die;
  open($fh2, ">read2/$fn_out") || die;
  while (<$fhin>) {
	if (/^@/) {
	  chomp; print $fh1 "$_/1\n"; print $fh2 "$_/2\n";
	  $_ = <$fhin>; chomp; # sequence
	  print $fh1 substr($_, $tl1, $size1-$tl1), "\n+\n";
	  print $fh2 substr($_, $size1+$tl2), "\n+\n";
	  <$fhin>; $_ = <$fhin>; chomp; # qualities
	  print $fh1 substr($_, $tl1, $size1-$tl1), "\n";
	  print $fh2 substr($_, $size1+$tl2), "\n";
	}
  }
  close($fhin); close($fh1); close($fh2);
}

#
# for Sanger's farm only
#

sub fastq2bfq {
  my %opts = (s=>1000000, e=>'fastq', d=>'', r=>'', a=>250);
  getopts('s:e:Ed:r:a:', \%opts);
  die qq(
Usage:   maq.pl fastq2bfq [-s nreads] [-e ext] <src_dir> <dst_dir>\n
Options: -s INT       number of reads per file [1000000]
         -e STR       extension of the read files [fastq]
         -d FILE      adapter sequence file [null]
         -r FILE      reference genome [hg18_male.bfa]
         -a INT       maximum insert size [$opts{a}]
         -E           the input in is Solexa's "export" format
\n) if (@ARGV < 2);
  my ($size, $ext) = ($opts{s}, $opts{e});
  my $is_export = (defined $opts{E})? 1 : 0;
  my ($src_dir, $dst_dir) = ($ARGV[0], $ARGV[1]);
  $dst_dir =~ s/\/$//;
  my ($fh_lst, $fh_pl);
  open($fh_lst, ">$dst_dir.lst") || die;
  open($fh_pl, ">$dst_dir.pl") || die;
  my $tmp_dir = "$dst_dir/tmp";
  my $cwd = getcwd;
  die("FATAL ERROR: source directory '$src_dir' does not exist!\n") unless (-d $src_dir);
  mkdir($dst_dir) unless (-d $dst_dir);
  mkdir($tmp_dir) unless (-d $tmp_dir);
  chdir($src_dir);
  my @list = `(find . -name "*.$ext" -follow; find . -name "*.$ext.gz" -follow)`; # get the list of files
  chdir($cwd);
  $size = $size >> 1 << 1;
  # run 'maq fastq2bfq'
  my $maq = gwhich("maq") || die("ERROR: Cannot find maq\n");
  my $fq_all2std = gwhich("fq_all2std.pl") || die("ERROR: Cannot find fq_all2std.pl\n");
  foreach (@list) {
	chomp; s/^\.\///;
	my $ori = $_;
	my $prog = (/\.gz$/i? 'zcat' : 'cat') . " $src_dir/$ori";
	if ($is_export) {
	  next if ($ori !~ /export/i);
	  $prog = "$prog | $fq_all2std export2sol | $maq sol2sanger - -";
	}
	s/\//-/g; s/\.gz$//i; s/\.$ext$//;
	my ($cur_size, $name);
	if ($is_export) {
	  # IMPORTANT: I do not know what single end reads look like.
	  #   possibly this part does not work well for SE reads.
	  if (/s_\d+_([12])_/) { # paired
		$cur_size = int($size/2);
		$name = "reads$1";
		s/(s_\d+_)[12]_/$1/;
	  }
	} else {
	  if (/read(s?)[12]/) { # paired
		$cur_size = int($size/2);
		$name = (/read(s?)1/i)? "reads1" : "reads2";
		s/read(s?)[12]//i;
	  } else {
		$cur_size = $size;
		$name = "reads.bfq";
	  }
	}
	s/--/-/g; s/^-//;
	$name = (($name =~ /[12]/)? "$_-PE" : "$_-SE") . ":$name";
	&run_cmd("$prog | $maq fastq2bfq -n $cur_size - $tmp_dir/$name");
  }
  # move files to separate directories
  my $n_jobs = 0;
  @list = `(cd $tmp_dir; ls)`;
  foreach (@list) {
	chomp;
	if (/(.*):(reads[12]?)\@(\d+)\.bfq$/) {
	  my $d = "$dst_dir/$1\@$3";
	  my $t = "$d/$2.bfq";
	  unless (/reads2\@\d+\.bfq/) {
		++$n_jobs;
		print $fh_lst "$d\n";
		mkdir($d);
	  }
	  move("$tmp_dir/$_", $t);
	}
  }
  system("rm -fr $tmp_dir");
  close($fh_lst);
  # generate configuration file for farm-run.pl
  $opts{d} = "-d ".abs_path($opts{d}) if ($opts{d});
  if ($opts{r}) {
	$opts{r} = abs_path($opts{r});
  } else {
	$opts{r} = '../../human_male.bfa';
  }
  print $fh_pl qq(\%fr_config =
(
  # The list of directories that should be processed. Do NOT change this.
  run_list=>'$dst_dir.lst',
  # number of LSF jobs at a time
  n_jobs=>$n_jobs,
  LSF_queue=>'long',
  LSF_resource=>'select[mem>800 && type==X86_64] rusage[mem=800]',
  # dir of the executables
  binary_path=>"$ENV{HOME}/lsf-prog",
  # dir of the scripts, if it is different from binary_path
  script_path=>"$ENV{HOME}/lsf-prog/scripts",
  # in each working directory, this function will be called.
  action=>\\&func
);

sub func
{
  my \$dir = shift; # the working directory
  # The current working dir is the one where *.bfq are staying.
  # usually the reference sequence is put two-level higher than
  # the current working dir.
  my \$maq_comm = "maq map -a $opts{a} $opts{d} $dst_dir.map $opts{r}";
  if (-f "reads.bfq") { # single end
    system("\$maq_comm reads.bfq");
  } elsif (-f "reads1.bfq" && ! -f "reads2.bfq") { # single end
    system("\$maq_comm reads1.bfq");
  } elsif (-f "reads1.bfq" && -f "reads2.bfq") { # paired end
    system("\$maq_comm reads1.bfq reads2.bfq");
  }
}
);
  close($fh_pl);
}

sub sv {
  my %opts = (i=>150, l=>35, q=>35, s=>7);
  getopts('s:i:l:q:', \%opts);
  die("
Usage:   maq.pl sv <in.mapview>\n
Options: -i INT    maximum insert size [$opts{i}]
         -l INT    average read length [$opts{l}]
         -q INT    minimum alternative mapping quality [$opts{q}]
         -s INT    minimum length of a region [$opts{s}]\n
") unless (@ARGV);
  my $d = $opts{i} - $opts{l};
  my ($begins, $beginc, $lasts, $lastc) = ('', -1, '', -1);
  my (@regs, %read, @reg_name);
  my @reg_seq;

  warn("-- read the mapview output\n");
  while (<>) {
	my @t = split;
	next if ($t[8] <= $opts{q});
	next if ($t[5] == 18 || $t[5] == 64 || $t[5] == 130);
	my $do_break = ($t[1] ne $lasts || $t[2] - $lastc > $d)? 1 : 0;
	if ($do_break) {
	  if ($lastc - $beginc > $opts{s}) { # skip short/unreliable regions
		my $k = @regs;
		my $flag = ($lastc - $beginc < $opts{i})? '*' : '.';
		push(@reg_name, "$begins\t$beginc\t$lastc\t$flag");
		my $p = \@{$regs[$k]};
		foreach (@reg_seq) {
		  push(@$p, $_);
		  my @s = split;
		  push(@{$read{$s[0]}}, $k);
		}
	  }
	  ($begins, $beginc) = @t[1..2];
	  @reg_seq = ();
	}
	$t[0] =~ s/\/[12]$//;
	push(@reg_seq, join(" ", @t[0..3,5,8,13]));
	($lasts, $lastc) = @t[1..2];
  }

  # build connections
  warn("-- link regions\n");
  my %link;
  foreach my $x (keys %read) {
	my $p = $read{$x};
	next if (@$p != 2);
	my $key = sprintf("%.10d %.10d", $p->[0], $p->[1]); # $p->[0] < $p->[1] always stands
	if (defined $link{$key}) {
	  ++$link{$key};
	} else {
	  $link{$key} = 1;
	}
  }

  # print connected regions
  warn("-- print result\n");
  foreach (reverse sort{$link{$a}<=>$link{$b}} keys %link) {
	my @s = split;
	my $x = $_;
	my @count;
	$count[$_] = 0 for (0..3);
	foreach my $y (@{$regs[$s[0]]}) {
	  my @t = split(" ", $y);
	  ++$count[($t[3] eq '+')? 0 : 1];
	}
	foreach my $y (@{$regs[$s[1]]}) {
	  my @t = split(" ", $y);
	  ++$count[($t[3] eq '+')? 2 : 3];
	}
	# infer flag
	my $flag = ($s[0] == $s[1])? 'LOP' : '';
	if (!$flag) {
	  my @t1 = split(/\s+/, $reg_name[$s[0]]);
	  my @t2 = split(/\s+/, $reg_name[$s[1]]);
	  if ($t1[0] ne $t2[0]) {
		$flag = 'DIF';
	  } elsif (($count[1] == 0 && 2*$count[2] < $count[3])
			   || ($count[2] == 0 && 2*$count[1] < $count[0])) {
		$flag = 'DEL';
	  } else {
		$flag = 'AMB';
	  }
	}
	print "$flag\t$link{$x}\t$reg_name[$s[0]]\t$count[0]\t$count[1]\t$reg_name[$s[1]]\t$count[2]\t$count[3]\n";
  }
}

#
# chrpt2snp command
#

sub chrpt2snp
{
	# calculate @mm
	my @mm;
	die("Usage: maq.pl chrpt2snp <chr_rpt.dbSNP>\n") unless (@ARGV);
	for (my $i = 0; $i != 0x10000; ++$i) {
		my ($x, $k) = (1, 0);
		for (my $l = 0; $l != 16; ++$l, $x <<= 1) {
			++$k if ($i & $x);
		}
		$mm[$i] = $k;
	}
	my $fh;
	open($fh, "| sort -k1,1 -k2,2n");
	while (<>) {
		next unless (/^\d+/ && /reference$/);
		my @t = split("\t", $_);
		next unless ($t[11] =~ /^\d+/); # no position
		my $vs = $mm[$t[16]&0xffff] + $mm[$t[16]>>16&0xffff];
		printf $fh ("chr$t[6]\t$t[11]\tN\tN\t%d\t%.3f\n", $vs*10, ($t[13] ne ' ')? $t[13] : 0.0);
	}
	close($fh);
}

#
# ucsc2snp command
#

sub ucsc2snp {
  warn("-- only one-basepair substitutions and indels will be retained.\n");
  my %conv = (AC=>'M', CA=>'M', AG=>'R', GA=>'R',
			  AT=>'W', TA=>'W', CG=>'S', GC=>'S',
			  CT=>'Y', TC=>'Y', GT=>'K', TG=>'K');
  my ($n_not2, $n_not1) = (0, 0);
  while (<>) {
	my @t = split;
	if ($t[9] !~ /-/) {
	  if (length($t[9]) != 3 || length($t[7]) != 1) {
		++$n_not1;
		next;
	  }
	}
	my ($a1, $a2) = split("/", $t[9]);
	my $SNP;
	if ($t[11] eq 'single') {
	  $SNP = $conv{"$a1$a2"};
	} else {
	  if ($t[7] eq '-') {
		$SNP = ($a1 ne '-')? $a1 : $a2;
	  } else {
		$SNP = '-';
	  }
	}
	my $qual = 0;
	if ($t[12] ne 'unknown') {
	  my @s = split(",", $t[12]);
	  $qual = @s * 10;
	}
	my $ref = (length($t[7]) == 1)? $t[7] : 'N';
	print "$t[1]\t$t[3]\t$ref\t$SNP\t$qual\t$t[13]\t$t[15]\n";
  }
  warn("-- discarded: ($n_not1, $n_not2)\n");
}

sub demo {
  my %opts = (N=>1000000, d=>'maqdemo');
  getopts('N:d:hs', \%opts);
  die("
Usage:   maq.pl demo [-N npairs] [-d outdir] [-h] <in.fa> <in.simudat>\n
Options: -N INT    number of read pairs [$opts{N}]
         -d DIR    output directory [$opts{d}]
         -s        single-end mode in alignment
         -h        haploid mode in simulation\n
") if (@ARGV < 2);
  my $peopt = (defined $opts{s})? '' : '-p';
  my $simuopt = "-N $opts{N}";
  $simuopt .= " -h" if (defined $opts{h});
  my $maq = gwhich("maq");
  my $maq_pl = gwhich("maq.pl");
  my $eval_pl = gwhich("maq_eval.pl");
  die("** 'maq', 'maq.pl' and 'maq_eval.pl' MUST be on the \$PATH\n") unless ($maq && $maq_pl && $eval_pl);
  &run_cmd("mkdir -p $opts{d}");
  &run_cmd("$maq simulate $simuopt $opts{d}/r1.fq $opts{d}/r2.fq $ARGV[0] $ARGV[1] > $opts{d}/true.snp");
  &run_cmd("$maq fasta2bfa $ARGV[0] $opts{d}/ref.bfa");
  &run_cmd("(cd $opts{d}; $maq_pl easyrun $peopt -d easyrun ref.bfa r1.fq r2.fq)");
  &run_cmd("(cd $opts{d}; $maq simustat easyrun/all.map > eval.simustat)");
  &run_cmd("(cd $opts{d}; $eval_pl sub -p eval.sub true.snp true.snp easyrun/cns.filter.snp)");
  &run_cmd("(cd $opts{d}; $eval_pl indelpe true.snp easyrun/cns.indelpe > eval.indelpe)") if ($peopt);
  &run_cmd("(cd $opts{d}; $eval_pl indelsoa true.snp easyrun/cns.indelse > eval.indelse)");
  warn("++ $opts{d}/easyrun/cns.filter.snp gives the SNPs that passes most of filters.\n");
  warn("++ $opts{d}/easyrun/cns.final.snp is the high-quality subset of the previous SNPs.\n");
  warn("++ $opts{d}/easyrun/mapcheck.txt gives some statistics about qualities.\n");
  warn("++ $opts{d}/true.snp contains all variants used in simulation.\n");
  warn("++ $opts{d}/eval.* give various benchmarks. Their formats will be documented later.\n");
}

#
#
#

sub indelpe {
  my %opts = (m=>6, n=>4);
  getopts('m:n:', \%opts);
  die("Usage: maq.pl indelpe [-m $opts{m}] [-n $opts{n}] <in.indelpe>\n") if (@ARGV == 0 && -t STDIN);
  my ($om, $on) = ($opts{m}, $opts{n});
  while (<>) {
	my @t = split;
	# calculate length of homopolymer tract
	$_ = $t[8];
	my $c = substr($_, 0, 1);
	/^($c+)/;
	my $n_hpt = length($1);
	$n_hpt = length($1) if (/^(${c}{3,}[^$c\s]{1,2}${c}{3,})/);
	$n_hpt = length($1) if (/^(${c}{4,}[^$c\s]{1,4}${c}{4,})/);
	# get rid of reads afftected by homopolymer tract
	my (@b, @e, @mm, @nb, @ne, @nmm);
	$_ = $t[10]; s/^@//; @b = split(',', $_);
	$_ = $t[11]; s/^@//; @e = split(',', $_);
	$_ = $t[12]; s/^@//; @mm = split(',', $_);
	for my $i (0 .. $#b) {
	  if ($b[$i] > $om && $e[$i] > $om && $e[$i] > $n_hpt + $on && $mm[$i] <= 2) {
		push(@nb, $b[$i]);
		push(@ne, $e[$i]);
		push(@nmm, $mm[$i]);
	  }
	}
	$t[3] = $t[5] + $t[6] + @nb;
	$t[9] = @nb;
	$t[10] = "@".join(",", @nb).",";
	$t[11] = "@".join(",", @ne).",";
	$t[12] = "@".join(",", @nmm).",";
	print join("\t", @t), "\n";
  }
}

#
# Usage
#

sub usage
{
	die qq(
Program: maq.pl (helper script for maq)
Version: $version
Contact: Heng Li <lh3\@sanger.ac.uk>

Usage:   maq.pl <command> [options] <input> [...]

Command: easyrun      simple pipeline for small dataset
         demo         demonstration of maq functionalities (for SE only)
         SNPfilter    filter SNPs
         statmap      extract statistics from the error output of 'maq map'
         chrpt2snp    convert dbSNP's chr_rpt file to .snp file
         ucsc2snp     convert UCSC's SNP dump to .snp file
         fastq2bfq    convert fastq in batch
         cat2pair     convert paired end reads format
         sv           call structural variations
         indelpe      post-processing indelpe output
\n);
}

#
# Other utilities
#

sub test_file
{
	my ($file) = @_;
	my $gzip = gwhich('gzip') || die;
	my ($is_gzip, $fh, $first);
	if (system("$gzip -l $file >/dev/null 2>&1")) {
		$is_gzip = 0;
		open($fh, $file);
	} else {
		$is_gzip = 1;
		open($fh, "$gzip -dc $file |");
	}
	read($fh, $first, 1);
	close($fh);
	return ($is_gzip, $first);
}

sub run_cmd
{
	my ($cmd) = @_;
	warn("-- CMD: $cmd\n");
	system("$cmd") && die("** fail to run command '$cmd'");
}

# the following codes are copied from treefam::generic

sub dirname
{
	my $prog = shift;
	my $cwd = getcwd;
	return $cwd if ($prog !~ /\//);
	$prog =~ s/\/[^\s\/]+$//g;
	return $prog;
}
sub which
{
	my $file = shift;
	my $path = (@_)? shift : $ENV{PATH};
	return if (!defined($path));
	foreach my $x (split(":", $path)) {
		$x =~ s/\/$//;
		return "$x/$file" if (-x "$x/$file" && -f "$x/$file");
	}
	return;
}
sub gwhich
{
	my $progname = shift;
	my $addtional_path = shift if (@_);
	my $dirname = &dirname($0);
	my $tmp;

	chomp($dirname);
	if (-x $progname && -f $progname) {
		return abs_path($progname);
	} elsif (defined($addtional_path) && ($tmp = &which($progname, $addtional_path))) {
		return $tmp; # lh3: Does it work? I will come back to this later
	} elsif (defined($dirname) && (-x "$dirname/$progname" && -f "$dirname/$progname")) {
		return abs_path("$dirname/$progname");
	} elsif (($tmp = &which($progname))) { # on the $PATH
		return $tmp;
	} else {
		warn("[gwhich] fail to find executable $progname anywhere.");
		return;
	}
}