annotate fastx_toolkit-0.0.6/src/fastq_quality_filter/fastq_quality_filter.c @ 3:997f5136985f draft default tip

Uploaded
author xilinxu
date Thu, 14 Aug 2014 04:52:17 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3
997f5136985f Uploaded
xilinxu
parents:
diff changeset
1 /*
997f5136985f Uploaded
xilinxu
parents:
diff changeset
2 FASTX-toolkit - FASTA/FASTQ preprocessing tools.
997f5136985f Uploaded
xilinxu
parents:
diff changeset
3 Copyright (C) 2009 A. Gordon (gordon@cshl.edu)
997f5136985f Uploaded
xilinxu
parents:
diff changeset
4
997f5136985f Uploaded
xilinxu
parents:
diff changeset
5 This program is free software: you can redistribute it and/or modify
997f5136985f Uploaded
xilinxu
parents:
diff changeset
6 it under the terms of the GNU Affero General Public License as
997f5136985f Uploaded
xilinxu
parents:
diff changeset
7 published by the Free Software Foundation, either version 3 of the
997f5136985f Uploaded
xilinxu
parents:
diff changeset
8 License, or (at your option) any later version.
997f5136985f Uploaded
xilinxu
parents:
diff changeset
9
997f5136985f Uploaded
xilinxu
parents:
diff changeset
10 This program is distributed in the hope that it will be useful,
997f5136985f Uploaded
xilinxu
parents:
diff changeset
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
997f5136985f Uploaded
xilinxu
parents:
diff changeset
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
997f5136985f Uploaded
xilinxu
parents:
diff changeset
13 GNU Affero General Public License for more details.
997f5136985f Uploaded
xilinxu
parents:
diff changeset
14
997f5136985f Uploaded
xilinxu
parents:
diff changeset
15 You should have received a copy of the GNU Affero General Public License
997f5136985f Uploaded
xilinxu
parents:
diff changeset
16 along with this program. If not, see <http://www.gnu.org/licenses/>.
997f5136985f Uploaded
xilinxu
parents:
diff changeset
17 */
997f5136985f Uploaded
xilinxu
parents:
diff changeset
18 #include <limits.h>
997f5136985f Uploaded
xilinxu
parents:
diff changeset
19 #include <stdio.h>
997f5136985f Uploaded
xilinxu
parents:
diff changeset
20 #include <stdlib.h>
997f5136985f Uploaded
xilinxu
parents:
diff changeset
21 #include <string.h>
997f5136985f Uploaded
xilinxu
parents:
diff changeset
22 #include <getopt.h>
997f5136985f Uploaded
xilinxu
parents:
diff changeset
23 #include <errno.h>
997f5136985f Uploaded
xilinxu
parents:
diff changeset
24 #include <err.h>
997f5136985f Uploaded
xilinxu
parents:
diff changeset
25
997f5136985f Uploaded
xilinxu
parents:
diff changeset
26 #include <config.h>
997f5136985f Uploaded
xilinxu
parents:
diff changeset
27
997f5136985f Uploaded
xilinxu
parents:
diff changeset
28 #include "fastx.h"
997f5136985f Uploaded
xilinxu
parents:
diff changeset
29 #include "fastx_args.h"
997f5136985f Uploaded
xilinxu
parents:
diff changeset
30
997f5136985f Uploaded
xilinxu
parents:
diff changeset
31 #define MAX_ADAPTER_LEN 100
997f5136985f Uploaded
xilinxu
parents:
diff changeset
32
997f5136985f Uploaded
xilinxu
parents:
diff changeset
33 const char* usage=
997f5136985f Uploaded
xilinxu
parents:
diff changeset
34 "usage: fastq_quality_filter [-h] [-v] [-q N] [-p N] [-z] [-i INFILE] [-o OUTFILE]\n" \
997f5136985f Uploaded
xilinxu
parents:
diff changeset
35 "\n" \
997f5136985f Uploaded
xilinxu
parents:
diff changeset
36 "version " VERSION "\n" \
997f5136985f Uploaded
xilinxu
parents:
diff changeset
37 " [-h] = This helpful help screen.\n" \
997f5136985f Uploaded
xilinxu
parents:
diff changeset
38 " [-q N] = Minimum quality score to keep.\n" \
997f5136985f Uploaded
xilinxu
parents:
diff changeset
39 " [-p N] = Minimum percent of bases that must have [-q] quality.\n" \
997f5136985f Uploaded
xilinxu
parents:
diff changeset
40 " [-z] = Compress output with GZIP.\n" \
997f5136985f Uploaded
xilinxu
parents:
diff changeset
41 " [-i INFILE] = FASTA/Q input file. default is STDIN.\n" \
997f5136985f Uploaded
xilinxu
parents:
diff changeset
42 " [-o OUTFILE] = FASTA/Q output file. default is STDOUT.\n" \
997f5136985f Uploaded
xilinxu
parents:
diff changeset
43 " [-v] = Verbose - report number of sequences.\n" \
997f5136985f Uploaded
xilinxu
parents:
diff changeset
44 " If [-o] is specified, report will be printed to STDOUT.\n" \
997f5136985f Uploaded
xilinxu
parents:
diff changeset
45 " If [-o] is not specified (and output goes to STDOUT),\n" \
997f5136985f Uploaded
xilinxu
parents:
diff changeset
46 " report will be printed to STDERR.\n" \
997f5136985f Uploaded
xilinxu
parents:
diff changeset
47 "\n";
997f5136985f Uploaded
xilinxu
parents:
diff changeset
48
997f5136985f Uploaded
xilinxu
parents:
diff changeset
49 #define DO_NOT_TRIM_LAST_BASE (0)
997f5136985f Uploaded
xilinxu
parents:
diff changeset
50
997f5136985f Uploaded
xilinxu
parents:
diff changeset
51 int min_quality=0;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
52 int min_percent=0;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
53
997f5136985f Uploaded
xilinxu
parents:
diff changeset
54 FASTX fastx;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
55
997f5136985f Uploaded
xilinxu
parents:
diff changeset
56 int parse_program_args(int __attribute__((unused)) optind, int optc, char* optarg)
997f5136985f Uploaded
xilinxu
parents:
diff changeset
57 {
997f5136985f Uploaded
xilinxu
parents:
diff changeset
58 switch(optc) {
997f5136985f Uploaded
xilinxu
parents:
diff changeset
59 case 'q':
997f5136985f Uploaded
xilinxu
parents:
diff changeset
60 if (optarg==NULL)
997f5136985f Uploaded
xilinxu
parents:
diff changeset
61 errx(1, "[-q] parameter requires an argument value");
997f5136985f Uploaded
xilinxu
parents:
diff changeset
62 min_quality = strtoul(optarg,NULL,10);
997f5136985f Uploaded
xilinxu
parents:
diff changeset
63 break;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
64
997f5136985f Uploaded
xilinxu
parents:
diff changeset
65 case 'p':
997f5136985f Uploaded
xilinxu
parents:
diff changeset
66 if (optarg==NULL)
997f5136985f Uploaded
xilinxu
parents:
diff changeset
67 errx(1, "[-l] parameter requires an argument value");
997f5136985f Uploaded
xilinxu
parents:
diff changeset
68 min_percent = strtoul(optarg,NULL,10);
997f5136985f Uploaded
xilinxu
parents:
diff changeset
69 if (min_percent<=0 || min_percent>100)
997f5136985f Uploaded
xilinxu
parents:
diff changeset
70 errx(1,"Invalid percent value (-p %s)", optarg);
997f5136985f Uploaded
xilinxu
parents:
diff changeset
71 break;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
72 default:
997f5136985f Uploaded
xilinxu
parents:
diff changeset
73 errx(1, __FILE__ ":%d: Unknown argument (%c)", __LINE__, optc ) ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
74 }
997f5136985f Uploaded
xilinxu
parents:
diff changeset
75 return 1;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
76 }
997f5136985f Uploaded
xilinxu
parents:
diff changeset
77
997f5136985f Uploaded
xilinxu
parents:
diff changeset
78 int get_index_of_nth_element(int *array, int array_size, int n)
997f5136985f Uploaded
xilinxu
parents:
diff changeset
79 {
997f5136985f Uploaded
xilinxu
parents:
diff changeset
80 int pos;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
81
997f5136985f Uploaded
xilinxu
parents:
diff changeset
82 //Find the first nono-empty index
997f5136985f Uploaded
xilinxu
parents:
diff changeset
83 pos = 0 ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
84 while ( pos < array_size && array[pos]==0 )
997f5136985f Uploaded
xilinxu
parents:
diff changeset
85 pos++;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
86
997f5136985f Uploaded
xilinxu
parents:
diff changeset
87 #if 0
997f5136985f Uploaded
xilinxu
parents:
diff changeset
88 fprintf(stderr,"n=%d\n", n);
997f5136985f Uploaded
xilinxu
parents:
diff changeset
89 for (i=0; i< array_size; i++) {
997f5136985f Uploaded
xilinxu
parents:
diff changeset
90 if (array[i] != 0)
997f5136985f Uploaded
xilinxu
parents:
diff changeset
91 fprintf(stderr, "[%d]=%d ", i + MIN_QUALITY_VALUE, array[i]) ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
92 }
997f5136985f Uploaded
xilinxu
parents:
diff changeset
93 fprintf(stderr,"\n");
997f5136985f Uploaded
xilinxu
parents:
diff changeset
94 #endif
997f5136985f Uploaded
xilinxu
parents:
diff changeset
95
997f5136985f Uploaded
xilinxu
parents:
diff changeset
96 if (pos == array_size)
997f5136985f Uploaded
xilinxu
parents:
diff changeset
97 errx(1,"bug: got empty array at %s:%d", __FILE__, __LINE__);
997f5136985f Uploaded
xilinxu
parents:
diff changeset
98
997f5136985f Uploaded
xilinxu
parents:
diff changeset
99 while (n > 0) {
997f5136985f Uploaded
xilinxu
parents:
diff changeset
100 if (array[pos] > n)
997f5136985f Uploaded
xilinxu
parents:
diff changeset
101 break;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
102 n -= array[pos];
997f5136985f Uploaded
xilinxu
parents:
diff changeset
103 pos++;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
104 while (array[pos]==0 && pos < array_size)
997f5136985f Uploaded
xilinxu
parents:
diff changeset
105 pos++;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
106 }
997f5136985f Uploaded
xilinxu
parents:
diff changeset
107 return pos;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
108 }
997f5136985f Uploaded
xilinxu
parents:
diff changeset
109
997f5136985f Uploaded
xilinxu
parents:
diff changeset
110 int get_percentile_quality(const FASTX *fastx, int percentile)
997f5136985f Uploaded
xilinxu
parents:
diff changeset
111 {
997f5136985f Uploaded
xilinxu
parents:
diff changeset
112 size_t i;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
113 int count=0;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
114 int quality_values[QUALITY_VALUES_RANGE];
997f5136985f Uploaded
xilinxu
parents:
diff changeset
115
997f5136985f Uploaded
xilinxu
parents:
diff changeset
116 memset(quality_values, 0, sizeof(quality_values));
997f5136985f Uploaded
xilinxu
parents:
diff changeset
117
997f5136985f Uploaded
xilinxu
parents:
diff changeset
118 for (i=0; i< strlen(fastx->nucleotides); i++) {
997f5136985f Uploaded
xilinxu
parents:
diff changeset
119 count++;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
120 quality_values[ fastx->quality[i] - MIN_QUALITY_VALUE ] ++ ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
121 }
997f5136985f Uploaded
xilinxu
parents:
diff changeset
122
997f5136985f Uploaded
xilinxu
parents:
diff changeset
123 i = get_index_of_nth_element(quality_values, QUALITY_VALUES_RANGE, (count * (100-percentile) / 100));
997f5136985f Uploaded
xilinxu
parents:
diff changeset
124
997f5136985f Uploaded
xilinxu
parents:
diff changeset
125 //printf(" n = %d, i = %d, i+MIN_QUAL_VALUE=%d\n",
997f5136985f Uploaded
xilinxu
parents:
diff changeset
126 // (count*(100-percentile)/100), i, i+MIN_QUALITY_VALUE) ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
127
997f5136985f Uploaded
xilinxu
parents:
diff changeset
128 return i + MIN_QUALITY_VALUE ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
129 }
997f5136985f Uploaded
xilinxu
parents:
diff changeset
130
997f5136985f Uploaded
xilinxu
parents:
diff changeset
131 int main(int argc, char* argv[])
997f5136985f Uploaded
xilinxu
parents:
diff changeset
132 {
997f5136985f Uploaded
xilinxu
parents:
diff changeset
133 fastx_parse_cmdline(argc, argv, "q:p:", parse_program_args);
997f5136985f Uploaded
xilinxu
parents:
diff changeset
134
997f5136985f Uploaded
xilinxu
parents:
diff changeset
135 fastx_init_reader(&fastx, get_input_filename(),
997f5136985f Uploaded
xilinxu
parents:
diff changeset
136 FASTQ_ONLY, ALLOW_N, REQUIRE_UPPERCASE);
997f5136985f Uploaded
xilinxu
parents:
diff changeset
137
997f5136985f Uploaded
xilinxu
parents:
diff changeset
138 fastx_init_writer(&fastx, get_output_filename(), OUTPUT_SAME_AS_INPUT, compress_output_flag());
997f5136985f Uploaded
xilinxu
parents:
diff changeset
139
997f5136985f Uploaded
xilinxu
parents:
diff changeset
140 while ( fastx_read_next_record(&fastx) ) {
997f5136985f Uploaded
xilinxu
parents:
diff changeset
141 #if 0
997f5136985f Uploaded
xilinxu
parents:
diff changeset
142 fprintf(stderr, "%s\n", fastx.nucleotides ) ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
143 for (i=0; i<strlen(fastx.nucleotides); i++) {
997f5136985f Uploaded
xilinxu
parents:
diff changeset
144 fprintf(stderr,"%d ", fastx.quality[i]);
997f5136985f Uploaded
xilinxu
parents:
diff changeset
145 }
997f5136985f Uploaded
xilinxu
parents:
diff changeset
146 fprintf(stderr,"\n");
997f5136985f Uploaded
xilinxu
parents:
diff changeset
147 #endif
997f5136985f Uploaded
xilinxu
parents:
diff changeset
148
997f5136985f Uploaded
xilinxu
parents:
diff changeset
149 int value = get_percentile_quality(&fastx, min_percent);
997f5136985f Uploaded
xilinxu
parents:
diff changeset
150
997f5136985f Uploaded
xilinxu
parents:
diff changeset
151 //fprintf(stderr, "value = %d\n\n", value ) ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
152
997f5136985f Uploaded
xilinxu
parents:
diff changeset
153
997f5136985f Uploaded
xilinxu
parents:
diff changeset
154 if (value >= min_quality) {
997f5136985f Uploaded
xilinxu
parents:
diff changeset
155 fastx_write_record(&fastx);
997f5136985f Uploaded
xilinxu
parents:
diff changeset
156 } else {
997f5136985f Uploaded
xilinxu
parents:
diff changeset
157 // fprintf(stderr, "%s\n", fastx.nucleotides ) ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
158 // fprintf(stderr, "value = %d\n", value ) ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
159 }
997f5136985f Uploaded
xilinxu
parents:
diff changeset
160 }
997f5136985f Uploaded
xilinxu
parents:
diff changeset
161
997f5136985f Uploaded
xilinxu
parents:
diff changeset
162 //
997f5136985f Uploaded
xilinxu
parents:
diff changeset
163 //Print verbose report
997f5136985f Uploaded
xilinxu
parents:
diff changeset
164 if ( verbose_flag() ) {
997f5136985f Uploaded
xilinxu
parents:
diff changeset
165 fprintf(get_report_file(), "Quality cut-off: %d\n", min_quality);
997f5136985f Uploaded
xilinxu
parents:
diff changeset
166 fprintf(get_report_file(), "Minimum percentage: %d\n", min_percent);
997f5136985f Uploaded
xilinxu
parents:
diff changeset
167
997f5136985f Uploaded
xilinxu
parents:
diff changeset
168 fprintf(get_report_file(), "Input: %zu reads.\n", num_input_reads(&fastx) ) ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
169 fprintf(get_report_file(), "Output: %zu reads.\n", num_output_reads(&fastx) ) ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
170
997f5136985f Uploaded
xilinxu
parents:
diff changeset
171 size_t discarded = num_input_reads(&fastx) - num_output_reads(&fastx) ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
172 fprintf(get_report_file(), "discarded %zu (%zu%%) low-quality reads.\n",
997f5136985f Uploaded
xilinxu
parents:
diff changeset
173 discarded, (discarded*100)/( num_input_reads(&fastx) ) ) ;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
174 }
997f5136985f Uploaded
xilinxu
parents:
diff changeset
175
997f5136985f Uploaded
xilinxu
parents:
diff changeset
176 return 0;
997f5136985f Uploaded
xilinxu
parents:
diff changeset
177 }