0
|
1 #!/usr/bin/perl -w
|
|
2
|
|
3 # Author: lh3
|
|
4
|
|
5 use strict;
|
|
6 use warnings;
|
|
7 use Getopt::Std;
|
|
8
|
|
9 my $version = '0.3.3';
|
|
10 &usage if (@ARGV < 1);
|
|
11
|
|
12 my $command = shift(@ARGV);
|
|
13 my %func = (showALEN=>\&showALEN, pileup2fq=>\&pileup2fq, varFilter=>\&varFilter,
|
|
14 unique=>\&unique, uniqcmp=>\&uniqcmp);
|
|
15
|
|
16 die("Unknown command \"$command\".\n") if (!defined($func{$command}));
|
|
17 &{$func{$command}};
|
|
18 exit(0);
|
|
19
|
|
20 #
|
|
21 # showALEN
|
|
22 #
|
|
23
|
|
24 sub showALEN {
|
|
25 die(qq/Usage: samtools.pl showALEN <in.sam>\n/) if (@ARGV == 0 && -t STDIN);
|
|
26 while (<>) {
|
|
27 my @t = split;
|
|
28 next if (/^\@/ || @t < 11);
|
|
29 my $l = 0;
|
|
30 $_ = $t[5];
|
|
31 s/(\d+)[MI]/$l+=$1/eg;
|
|
32 print join("\t", @t[0..5]), "\t$l\t", join("\t", @t[6..$#t]), "\n";
|
|
33 }
|
|
34 }
|
|
35
|
|
36 #
|
|
37 # varFilter
|
|
38 #
|
|
39
|
|
40 sub varFilter {
|
|
41 my %opts = (d=>3, D=>100, l=>30, Q=>25, q=>10, G=>25, s=>100, w=>10, W=>10, N=>2, p=>undef);
|
|
42 getopts('pq:d:D:l:Q:w:W:N:G:', \%opts);
|
|
43 die(qq/
|
|
44 Usage: samtools.pl varFilter [options] <in.cns-pileup>
|
|
45
|
|
46 Options: -Q INT minimum RMS mapping quality for SNPs [$opts{Q}]
|
|
47 -q INT minimum RMS mapping quality for gaps [$opts{q}]
|
|
48 -d INT minimum read depth [$opts{d}]
|
|
49 -D INT maximum read depth [$opts{D}]
|
|
50
|
|
51 -G INT min indel score for nearby SNP filtering [$opts{G}]
|
|
52 -w INT SNP within INT bp around a gap to be filtered [$opts{w}]
|
|
53
|
|
54 -W INT window size for filtering dense SNPs [$opts{W}]
|
|
55 -N INT max number of SNPs in a window [$opts{N}]
|
|
56
|
|
57 -l INT window size for filtering adjacent gaps [$opts{l}]
|
|
58
|
|
59 -p print filtered variants
|
|
60 \n/) if (@ARGV == 0 && -t STDIN);
|
|
61
|
|
62 # calculate the window size
|
|
63 my ($ol, $ow, $oW) = ($opts{l}, $opts{w}, $opts{W});
|
|
64 my $max_dist = $ol > $ow? $ol : $ow;
|
|
65 $max_dist = $oW if ($max_dist < $oW);
|
|
66 # the core loop
|
|
67 my @staging; # (indel_filtering_score, flt_tag)
|
|
68 while (<>) {
|
|
69 my @t = split;
|
|
70 next if (uc($t[2]) eq uc($t[3]) || $t[3] eq '*/*'); # skip non-var sites
|
|
71 # clear the out-of-range elements
|
|
72 while (@staging) {
|
|
73 last if ($staging[0][2] eq $t[0] && $staging[0][3] + $max_dist >= $t[1]);
|
|
74 varFilter_aux(shift(@staging), $opts{p}); # calling a function is a bit slower, not much
|
|
75 }
|
|
76 my ($flt, $score) = (0, -1);
|
|
77 # first a simple filter
|
|
78 if ($t[7] < $opts{d}) {
|
|
79 $flt = 2;
|
|
80 } elsif ($t[7] > $opts{D}) {
|
|
81 $flt = 3;
|
|
82 }
|
|
83 # site dependent filters
|
|
84 if ($flt == 0) {
|
|
85 if ($t[2] eq '*') { # an indel
|
|
86 $flt = 1 if ($t[6] < $opts{q});
|
|
87 # filtering SNPs
|
|
88 if ($t[5] >= $opts{G}) {
|
|
89 for my $x (@staging) {
|
|
90 next if ($x->[0] >= 0 || $x->[3] + $ow < $t[1]);
|
|
91 $x->[1] = 5 if ($x->[1] == 0);
|
|
92 }
|
|
93 }
|
|
94 # calculate the filtering score (different from indel quality)
|
|
95 $score = $t[5];
|
|
96 $score += $opts{s} * $t[10] if ($t[8] ne '*');
|
|
97 $score += $opts{s} * $t[11] if ($t[9] ne '*');
|
|
98 # check the staging list for indel filtering
|
|
99 for my $x (@staging) {
|
|
100 next if ($x->[0] < 0 || $x->[3] + $ol < $t[1]);
|
|
101 if ($x->[0] < $score) {
|
|
102 $x->[1] = 6;
|
|
103 } else {
|
|
104 $flt = 6; last;
|
|
105 }
|
|
106 }
|
|
107 } else { # a SNP
|
|
108 $flt = 1 if ($t[6] < $opts{Q});
|
|
109 # check adjacent SNPs
|
|
110 my $k = 1;
|
|
111 for my $x (@staging) {
|
|
112 ++$k if ($x->[0] < 0 && $x->[3] + $oW >= $t[1] && ($x->[1] == 0 || $x->[1] == 4 || $x->[1] == 5));
|
|
113 }
|
|
114 # filtering is necessary
|
|
115 if ($k > $opts{N}) {
|
|
116 $flt = 4;
|
|
117 for my $x (@staging) {
|
|
118 $x->[1] = 4 if ($x->[0] < 0 && $x->[3] + $oW >= $t[1] && $x->[1] == 0);
|
|
119 }
|
|
120 } else { # then check gap filter
|
|
121 for my $x (@staging) {
|
|
122 next if ($x->[0] < 0 || $x->[3] + $ow < $t[1]);
|
|
123 if ($x->[0] >= $opts{G}) {
|
|
124 $flt = 5; last;
|
|
125 }
|
|
126 }
|
|
127 }
|
|
128 }
|
|
129 }
|
|
130 push(@staging, [$score, $flt, @t]);
|
|
131 }
|
|
132 # output the last few elements in the staging list
|
|
133 while (@staging) {
|
|
134 varFilter_aux(shift @staging, $opts{p});
|
|
135 }
|
|
136 }
|
|
137
|
|
138 sub varFilter_aux {
|
|
139 my ($first, $is_print) = @_;
|
|
140 if ($first->[1] == 0) {
|
|
141 print join("\t", @$first[2 .. @$first-1]), "\n";
|
|
142 } elsif ($is_print) {
|
|
143 print STDERR join("\t", substr("UQdDWGgX", $first->[1], 1), @$first[2 .. @$first-1]), "\n";
|
|
144 }
|
|
145 }
|
|
146
|
|
147 #
|
|
148 # pileup2fq
|
|
149 #
|
|
150
|
|
151 sub pileup2fq {
|
|
152 my %opts = (d=>3, D=>255, Q=>25, G=>25, l=>10);
|
|
153 getopts('d:D:Q:G:l:', \%opts);
|
|
154 die(qq/
|
|
155 Usage: samtools.pl pileup2fq [options] <in.cns-pileup>
|
|
156
|
|
157 Options: -d INT minimum depth [$opts{d}]
|
|
158 -D INT maximum depth [$opts{D}]
|
|
159 -Q INT min RMS mapQ [$opts{Q}]
|
|
160 -G INT minimum indel score [$opts{G}]
|
|
161 -l INT indel filter winsize [$opts{l}]\n
|
|
162 /) if (@ARGV == 0 && -t STDIN);
|
|
163
|
|
164 my ($last_chr, $seq, $qual, @gaps, $last_pos);
|
|
165 my $_Q = $opts{Q};
|
|
166 my $_d = $opts{d};
|
|
167 my $_D = $opts{D};
|
|
168
|
|
169 $last_chr = '';
|
|
170 while (<>) {
|
|
171 my @t = split;
|
|
172 if ($last_chr ne $t[0]) {
|
|
173 &p2q_post_process($last_chr, \$seq, \$qual, \@gaps, $opts{l}) if ($last_chr);
|
|
174 $last_chr = $t[0];
|
|
175 $last_pos = 0;
|
|
176 $seq = ''; $qual = '';
|
|
177 @gaps = ();
|
|
178 }
|
|
179 if ($t[1] - $last_pos != 1) {
|
|
180 $seq .= 'n' x ($t[1] - $last_pos - 1);
|
|
181 $qual .= '!' x ($t[1] - $last_pos - 1);
|
|
182 }
|
|
183 if ($t[2] eq '*') {
|
|
184 push(@gaps, $t[1]) if ($t[5] >= $opts{G});
|
|
185 } else {
|
|
186 $seq .= ($t[6] >= $_Q && $t[7] >= $_d && $t[7] <= $_D)? uc($t[3]) : lc($t[3]);
|
|
187 my $q = $t[4] + 33;
|
|
188 $q = 126 if ($q > 126);
|
|
189 $qual .= chr($q);
|
|
190 }
|
|
191 $last_pos = $t[1];
|
|
192 }
|
|
193 &p2q_post_process($last_chr, \$seq, \$qual, \@gaps, $opts{l});
|
|
194 }
|
|
195
|
|
196 sub p2q_post_process {
|
|
197 my ($chr, $seq, $qual, $gaps, $l) = @_;
|
|
198 &p2q_filter_gaps($seq, $gaps, $l);
|
|
199 print "\@$chr\n"; &p2q_print_str($seq);
|
|
200 print "+\n"; &p2q_print_str($qual);
|
|
201 }
|
|
202
|
|
203 sub p2q_filter_gaps {
|
|
204 my ($seq, $gaps, $l) = @_;
|
|
205 for my $g (@$gaps) {
|
|
206 my $x = $g > $l? $g - $l : 0;
|
|
207 substr($$seq, $x, $l + $l) = lc(substr($$seq, $x, $l + $l));
|
|
208 }
|
|
209 }
|
|
210
|
|
211 sub p2q_print_str {
|
|
212 my ($s) = @_;
|
|
213 my $l = length($$s);
|
|
214 for (my $i = 0; $i < $l; $i += 60) {
|
|
215 print substr($$s, $i, 60), "\n";
|
|
216 }
|
|
217 }
|
|
218
|
|
219 #
|
|
220 # unique
|
|
221 #
|
|
222
|
|
223 sub unique {
|
|
224 my %opts = (f=>250.0, q=>5, r=>2, a=>1, b=>3);
|
|
225 getopts('Qf:q:r:a:b:', \%opts);
|
|
226 die("Usage: samtools.pl unique [-f $opts{f}] <in.sam>\n") if (@ARGV == 0 && -t STDIN);
|
|
227 my $last = '';
|
|
228 my $recal_Q = !defined($opts{Q});
|
|
229 my @a;
|
|
230 while (<>) {
|
|
231 my $score = -1;
|
|
232 print $_ if (/^\@/);
|
|
233 $score = $1 if (/AS:i:(\d+)/);
|
|
234 my @t = split("\t");
|
|
235 next if (@t < 11);
|
|
236 if ($score < 0) { # AS tag is unavailable
|
|
237 my $cigar = $t[5];
|
|
238 my ($mm, $go, $ge) = (0, 0, 0);
|
|
239 $cigar =~ s/(\d+)[ID]/++$go,$ge+=$1/eg;
|
|
240 $cigar = $t[5];
|
|
241 $cigar =~ s/(\d+)M/$mm+=$1/eg;
|
|
242 $score = $mm * $opts{a} - $go * $opts{q} - $ge * $opts{r}; # no mismatches...
|
|
243 }
|
|
244 $score = 1 if ($score < 1);
|
|
245 if ($t[0] ne $last) {
|
|
246 &unique_aux(\@a, $opts{f}, $recal_Q) if (@a);
|
|
247 $last = $t[0];
|
|
248 }
|
|
249 push(@a, [$score, \@t]);
|
|
250 }
|
|
251 &unique_aux(\@a, $opts{f}, $recal_Q) if (@a);
|
|
252 }
|
|
253
|
|
254 sub unique_aux {
|
|
255 my ($a, $fac, $is_recal) = @_;
|
|
256 my ($max, $max2, $max_i) = (0, 0, -1);
|
|
257 for (my $i = 0; $i < @$a; ++$i) {
|
|
258 if ($a->[$i][0] > $max) {
|
|
259 $max2 = $max; $max = $a->[$i][0]; $max_i = $i;
|
|
260 } elsif ($a->[$i][0] > $max2) {
|
|
261 $max2 = $a->[$i][0];
|
|
262 }
|
|
263 }
|
|
264 if ($is_recal) {
|
|
265 my $q = int($fac * ($max - $max2) / $max + .499);
|
|
266 $q = 250 if ($q > 250);
|
|
267 $a->[$max_i][1][4] = $q < 250? $q : 250;
|
|
268 }
|
|
269 print join("\t", @{$a->[$max_i][1]});
|
|
270 @$a = ();
|
|
271 }
|
|
272
|
|
273 #
|
|
274 # uniqcmp: compare two SAM files
|
|
275 #
|
|
276
|
|
277 sub uniqcmp {
|
|
278 my %opts = (q=>10, s=>100);
|
|
279 getopts('pq:s:', \%opts);
|
|
280 die("Usage: samtools.pl uniqcmp <in1.sam> <in2.sam>\n") if (@ARGV < 2);
|
|
281 my ($fh, %a);
|
|
282 warn("[uniqcmp] read the first file...\n");
|
|
283 &uniqcmp_aux($ARGV[0], \%a, 0);
|
|
284 warn("[uniqcmp] read the second file...\n");
|
|
285 &uniqcmp_aux($ARGV[1], \%a, 1);
|
|
286 warn("[uniqcmp] stats...\n");
|
|
287 my @cnt;
|
|
288 $cnt[$_] = 0 for (0..9);
|
|
289 for my $x (keys %a) {
|
|
290 my $p = $a{$x};
|
|
291 my $z;
|
|
292 if (defined($p->[0]) && defined($p->[1])) {
|
|
293 $z = ($p->[0][0] == $p->[1][0] && $p->[0][1] eq $p->[1][1] && abs($p->[0][2] - $p->[1][2]) < $opts{s})? 0 : 1;
|
|
294 if ($p->[0][3] >= $opts{q} && $p->[1][3] >= $opts{q}) {
|
|
295 ++$cnt[$z*3+0];
|
|
296 } elsif ($p->[0][3] >= $opts{q}) {
|
|
297 ++$cnt[$z*3+1];
|
|
298 } elsif ($p->[1][3] >= $opts{q}) {
|
|
299 ++$cnt[$z*3+2];
|
|
300 }
|
|
301 print STDERR "$x\t$p->[0][1]:$p->[0][2]\t$p->[0][3]\t$p->[0][4]\t$p->[1][1]:$p->[1][2]\t$p->[1][3]\t$p->[1][4]\t",
|
|
302 $p->[0][5]-$p->[1][5], "\n" if ($z && defined($opts{p}) && ($p->[0][3] >= $opts{q} || $p->[1][3] >= $opts{q}));
|
|
303 } elsif (defined($p->[0])) {
|
|
304 ++$cnt[$p->[0][3]>=$opts{q}? 6 : 7];
|
|
305 print STDERR "$x\t$p->[0][1]:$p->[0][2]\t$p->[0][3]\t$p->[0][4]\t*\t0\t*\t",
|
|
306 $p->[0][5], "\n" if (defined($opts{p}) && $p->[0][3] >= $opts{q});
|
|
307 } else {
|
|
308 print STDERR "$x\t*\t0\t*\t$p->[1][1]:$p->[1][2]\t$p->[1][3]\t$p->[1][4]\t",
|
|
309 -$p->[1][5], "\n" if (defined($opts{p}) && $p->[1][3] >= $opts{q});
|
|
310 ++$cnt[$p->[1][3]>=$opts{q}? 8 : 9];
|
|
311 }
|
|
312 }
|
|
313 print "Consistent (high, high): $cnt[0]\n";
|
|
314 print "Consistent (high, low ): $cnt[1]\n";
|
|
315 print "Consistent (low , high): $cnt[2]\n";
|
|
316 print "Inconsistent (high, high): $cnt[3]\n";
|
|
317 print "Inconsistent (high, low ): $cnt[4]\n";
|
|
318 print "Inconsistent (low , high): $cnt[5]\n";
|
|
319 print "Second missing (high): $cnt[6]\n";
|
|
320 print "Second missing (low ): $cnt[7]\n";
|
|
321 print "First missing (high): $cnt[8]\n";
|
|
322 print "First missing (low ): $cnt[9]\n";
|
|
323 }
|
|
324
|
|
325 sub uniqcmp_aux {
|
|
326 my ($fn, $a, $which) = @_;
|
|
327 my $fh;
|
|
328 $fn = "samtools view $fn |" if ($fn =~ /\.bam/);
|
|
329 open($fh, $fn) || die;
|
|
330 while (<$fh>) {
|
|
331 my @t = split;
|
|
332 next if (@t < 11);
|
|
333 # my $l = ($t[5] =~ /^(\d+)S/)? $1 : 0;
|
|
334 my $l = 0;
|
|
335 my ($x, $nm) = (0, 0);
|
|
336 $nm = $1 if (/NM:i:(\d+)/);
|
|
337 $_ = $t[5];
|
|
338 s/(\d+)[MI]/$x+=$1/eg;
|
|
339 @{$a->{$t[0]}[$which]} = (($t[1]&0x10)? 1 : 0, $t[2], $t[3]-$l, $t[4], "$x:$nm", $x - 4 * $nm);
|
|
340 }
|
|
341 close($fh);
|
|
342 }
|
|
343
|
|
344 #
|
|
345 # Usage
|
|
346 #
|
|
347
|
|
348 sub usage {
|
|
349 die(qq/
|
|
350 Program: samtools.pl (helper script for SAMtools)
|
|
351 Version: $version
|
|
352 Contact: Heng Li <lh3\@sanger.ac.uk>\n
|
|
353 Usage: samtools.pl <command> [<arguments>]\n
|
|
354 Command: varFilter filtering SNPs and short indels
|
|
355 pileup2fq generate fastq from `pileup -c'
|
|
356 showALEN print alignment length (ALEN) following CIGAR
|
|
357 \n/);
|
|
358 }
|