annotate clustalomega/clustal-omega-1.0.2/src/squid/a2m.c @ 1:bc707542e5de

Uploaded
author clustalomega
date Thu, 21 Jul 2011 13:35:08 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
1 /*****************************************************************
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
2 * SQUID - a library of functions for biological sequence analysis
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
3 * Copyright (C) 1992-2002 Washington University School of Medicine
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
4 *
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
5 * This source code is freely distributed under the terms of the
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
6 * GNU General Public License. See the files COPYRIGHT and LICENSE
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
7 * for details.
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
8 *****************************************************************/
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
9
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
10 /* a2m.c
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
11 *
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
12 * reading/writing A2M (aligned FASTA) files.
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
13 *
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
14 * RCS $Id: a2m.c 242 2011-05-27 14:04:21Z andreas $ (Original squid RCS Id: a2m.c,v 1.1 1999/07/15 22:26:40 eddy Exp)
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
15 */
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
16
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
17 #include <stdio.h>
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
18 #include <stdlib.h>
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
19 #include <string.h>
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
20 #include "squid.h"
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
21 #include "msa.h"
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
22
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
23 /* Function: ReadA2M()
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
24 * Date: SRE, Sun Jun 6 17:11:29 1999 [bus from Madison 1999 worm mtg]
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
25 *
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
26 * Purpose: Parse an alignment read from an open A2M format
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
27 * alignment file. A2M is a single alignment format.
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
28 * Return the alignment, or NULL if we've already
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
29 * read the alignment.
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
30 *
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
31 * Args: afp - open alignment file
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
32 *
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
33 * Returns: MSA * - an alignment object.
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
34 * Caller responsible for an MSAFree()
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
35 */
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
36 MSA *
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
37 ReadA2M(MSAFILE *afp)
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
38 {
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
39 MSA *msa;
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
40 char *buf;
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
41 char *name;
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
42 char *desc;
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
43 char *seq;
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
44 int idx;
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
45 int len1, len2;
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
46
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
47 if (feof(afp->f)) return NULL;
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
48
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
49 name = NULL;
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
50 msa = MSAAlloc(10, 0);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
51 idx = 0;
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
52 while ((buf = MSAFileGetLine(afp)) != NULL)
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
53 {
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
54 if (*buf == '>')
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
55 {
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
56 buf++; /* skip the '>' */
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
57 if ((name = sre_strtok(&buf, WHITESPACE, &len1)) == NULL)
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
58 Die("Blank name in A2M file %s (line %d)\n", afp->fname, afp->linenumber);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
59 desc = sre_strtok(&buf, "\n", &len2);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
60
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
61 idx = GKIStoreKey(msa->index, name);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
62 if (idx >= msa->nseqalloc) MSAExpand(msa);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
63
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
64 msa->sqname[idx] = sre_strdup(name, len1);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
65 if (desc != NULL) MSASetSeqDescription(msa, idx, desc);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
66 msa->nseq++;
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
67 }
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
68 else if (name != NULL)
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
69 {
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
70 if ((seq = sre_strtok(&buf, WHITESPACE, &len1)) == NULL) continue;
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
71 msa->sqlen[idx] = sre_strcat(&(msa->aseq[idx]), msa->sqlen[idx], seq, len1);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
72 }
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
73 }
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
74 if (name == NULL) { MSAFree(msa); return NULL; }
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
75
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
76 MSAVerifyParse(msa);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
77 return msa;
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
78 }
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
79
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
80
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
81 /* Function: WriteA2M()
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
82 * Date: SRE, Sun Jun 6 17:40:35 1999 [bus from Madison, 1999 worm mtg]
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
83 *
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
84 * Purpose: Write an "aligned FASTA" (aka a2m, to UCSC) formatted
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
85 * alignment.
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
86 *
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
87 * Args: fp - open FILE to write to.
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
88 * msa - alignment to write
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
89 *
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
90 * Returns: void
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
91 */
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
92 void
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
93 #ifdef CLUSTALO
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
94 WriteA2M(FILE *fp, MSA *msa, int vienna)
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
95 #else
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
96 WriteA2M(FILE *fp, MSA *msa)
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
97 #endif
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
98 {
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
99 int idx; /* sequence index */
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
100 int pos; /* position in sequence */
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
101 char buf[64]; /* buffer for individual lines */
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
102 int cpl = 60; /* char per line; must be < 64 unless buf is bigger */
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
103
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
104 buf[cpl] = '\0';
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
105 for (idx = 0; idx < msa->nseq; idx++)
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
106 {
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
107 #ifdef CLUSTALO
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
108 /* most fasta sequences don't have a description, which
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
109 * leads to a trailing white space in the original code
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
110 */
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
111 fprintf(fp, ">%s", msa->sqname[idx]);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
112
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
113 if (msa->sqdesc != NULL && msa->sqdesc[idx] != NULL && !vienna) {
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
114 fprintf(fp, " %s", msa->sqdesc[idx]);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
115 }
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
116 fprintf(fp, "\n");
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
117 #else
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
118 fprintf(fp, ">%s %s\n",
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
119 msa->sqname[idx],
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
120 (msa->sqdesc != NULL && msa->sqdesc[idx] != NULL) ? msa->sqdesc[idx] : "");
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
121 #endif
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
122 for (pos = 0; pos < msa->alen; pos+=cpl)
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
123 {
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
124 strncpy(buf, &(msa->aseq[idx][pos]), cpl);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
125 if (vienna)
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
126 fprintf(fp, "%s", buf);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
127 else
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
128 fprintf(fp, "%s\n", buf);
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
129 }
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
130 if (vienna)
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
131 fprintf(fp, "\n");
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
132
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
133 }
bc707542e5de Uploaded
clustalomega
parents:
diff changeset
134 }