annotate mrsfast-2.3.0.2/RefGenome.c @ 0:ec628ba33878 default tip

Uploaded source code for mrsFAST
author calkan
date Tue, 21 Feb 2012 10:39:28 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
1 /*
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
2 * Copyright (c) <2008 - 2009>, University of Washington, Simon Fraser University
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
3 * All rights reserved.
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
4 *
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
5 * Redistribution and use in source and binary forms, with or without modification,
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
6 * are permitted provided that the following conditions are met:
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
7 *
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
8 * Redistributions of source code must retain the above copyright notice, this list
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
9 * of conditions and the following disclaimer.
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
10 * - Redistributions in binary form must reproduce the above copyright notice, this
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
11 * list of conditions and the following disclaimer in the documentation and/or other
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
12 * materials provided with the distribution.
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
13 * - Neither the name of the <ORGANIZATION> nor the names of its contributors may be
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
14 * used to endorse or promote products derived from this software without specific
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
15 * prior written permission.
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
16 *
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
17 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
18 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
19 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
20 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
21 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
22 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
23 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
24 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
25 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
26 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
27 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
28 */
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
29
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
30 /*
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
31 * Author : Faraz Hach
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
32 * Email : fhach AT cs DOT sfu
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
33 * Last Update : 2009-12-08
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
34 */
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
35
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
36
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
37 #include <stdio.h>
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
38 #include <stdlib.h>
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
39 #include <string.h>
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
40 #include <ctype.h>
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
41 #include "Common.h"
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
42 #include "RefGenome.h"
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
43
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
44 FILE *_rg_fp;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
45 char *_rg_gen;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
46 char *_rg_name;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
47 int _rg_offset;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
48 int _rg_contGen;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
49
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
50 /**********************************************/
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
51 int initLoadingRefGenome(char *fileName)
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
52 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
53 char ch;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
54 _rg_fp = fileOpen (fileName, "r");
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
55 if (fscanf(_rg_fp, "%c", &ch))
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
56 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
57 if (ch == '>')
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
58 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
59 _rg_contGen = 0;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
60 _rg_offset = 0;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
61 _rg_gen = getMem(CONTIG_MAX_SIZE);
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
62 _rg_name = getMem(CONTIG_NAME_SIZE);
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
63 return 1;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
64 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
65 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
66 return 0;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
67 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
68 /**********************************************/
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
69 void finalizeLoadingRefGenome()
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
70 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
71 freeMem(_rg_gen, CONTIG_MAX_SIZE);
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
72 freeMem(_rg_name, CONTIG_NAME_SIZE);
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
73 fclose(_rg_fp);
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
74 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
75 /**********************************************/
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
76 int loadRefGenome(char **refGen, char **refGenName, int *refGenOff)
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
77 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
78 char ch;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
79 int i;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
80 int returnVal = 0;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
81 int actualSize=0;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
82 int size;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
83 char *tmp;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
84
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
85 // New Conting
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
86 if (!_rg_contGen)
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
87 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
88 size = 0;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
89 tmp = fgets(_rg_name, SEQ_MAX_LENGTH, _rg_fp);
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
90 int k;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
91 for (k=0; k<strlen(_rg_name);k++)
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
92 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
93 if (_rg_name[k] == ' ')
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
94 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
95 _rg_name[k]='\0';
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
96 break;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
97 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
98 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
99
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
100 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
101 else
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
102 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
103 size=strlen(_rg_gen);
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
104 for( i = 0 ; i < CONTIG_OVERLAP ; i++ )
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
105 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
106
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
107 _rg_gen[i] = _rg_gen[size-CONTIG_OVERLAP+i];
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
108 if (_rg_gen[i] != 'N')
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
109 actualSize++;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
110 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
111 size = CONTIG_OVERLAP;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
112 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
113 while( fscanf(_rg_fp, "%c", &ch) > 0 )
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
114 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
115 if (ch == '>')
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
116 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
117 _rg_contGen = 0;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
118 returnVal = 1;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
119 break;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
120 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
121 else if (!isspace(ch))
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
122 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
123 ch = toupper(ch);
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
124 _rg_gen[size++] = ch;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
125 if (ch != 'N')
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
126 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
127 actualSize++;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
128 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
129 if (actualSize == CONTIG_SIZE || size == CONTIG_MAX_SIZE)
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
130 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
131 _rg_contGen = 1;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
132 returnVal=1;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
133 break;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
134 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
135 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
136
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
137 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
138
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
139 _rg_gen[size] = '\0';
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
140 for (i=strlen(_rg_name)-1; i >= 0; i--)
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
141 if (!isspace(_rg_name[i]))
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
142 break;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
143 _rg_name[i+1] = '\0';
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
144
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
145 *refGenOff = _rg_offset;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
146 *refGenName = _rg_name;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
147 *refGen = _rg_gen;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
148
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
149 if (_rg_contGen == 1)
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
150 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
151 _rg_offset += size-CONTIG_OVERLAP;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
152 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
153 else
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
154 {
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
155 _rg_offset = 0;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
156 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
157
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
158
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
159 return returnVal;
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
160 }
ec628ba33878 Uploaded source code for mrsFAST
calkan
parents:
diff changeset
161 /**********************************************/