1 | /* data structure for each file format and sequences */ |
---|
2 | |
---|
3 | #ifndef FILEBUFFER_H |
---|
4 | #include <FileBuffer.h> |
---|
5 | #endif |
---|
6 | |
---|
7 | #define INITSEQ 6000 |
---|
8 | #define RNA 0 |
---|
9 | #define NONRNA 1 |
---|
10 | #define EQ 1 |
---|
11 | #define NONEQ 0 |
---|
12 | |
---|
13 | #define GENBANK 'g' |
---|
14 | #define MACKE 'm' |
---|
15 | #define PROTEIN 't' |
---|
16 | #define PHYLIP 'y' |
---|
17 | #define PHYLIP2 '2' |
---|
18 | #define PAUP 'p' |
---|
19 | #define EMBL 'e' |
---|
20 | #define GCG 'c' |
---|
21 | #define PRINTABLE 'r' |
---|
22 | #define ALMA 'l' |
---|
23 | #define NBRF 'n' |
---|
24 | #define STADEN 's' |
---|
25 | |
---|
26 | #define UNKNOWN (-1) |
---|
27 | #define AUTHOR 1 |
---|
28 | #define JOURNAL 2 |
---|
29 | #define TITLE 3 |
---|
30 | #define STANDARD 4 |
---|
31 | #define PROCESS 5 |
---|
32 | #define REF 6 |
---|
33 | #define ALL 0 |
---|
34 | |
---|
35 | #define SEQLINE 60 |
---|
36 | /* GenBank data structure */ |
---|
37 | typedef struct { |
---|
38 | char *ref; |
---|
39 | char *author; |
---|
40 | char *title; |
---|
41 | char *journal; |
---|
42 | char *standard; |
---|
43 | } Reference; |
---|
44 | |
---|
45 | typedef struct { |
---|
46 | int exist; |
---|
47 | char *source; |
---|
48 | char *cc; |
---|
49 | char *formname; |
---|
50 | char *nickname; |
---|
51 | char *commname; |
---|
52 | char *hostorg; |
---|
53 | } Oinf; |
---|
54 | |
---|
55 | typedef struct { |
---|
56 | int exist; |
---|
57 | char *RDPid; |
---|
58 | char *gbkentry; |
---|
59 | char *methods; |
---|
60 | char comp5; /* yes or no, y/n */ |
---|
61 | char comp3; /* yes or no, y/n */ |
---|
62 | } Sinf; |
---|
63 | |
---|
64 | typedef struct { |
---|
65 | Oinf orginf; |
---|
66 | Sinf seqinf; |
---|
67 | char *others; |
---|
68 | } Comment; |
---|
69 | |
---|
70 | typedef struct { |
---|
71 | char *locus; |
---|
72 | char *definition; |
---|
73 | char *accession; |
---|
74 | char *keywords; |
---|
75 | char *source; |
---|
76 | char *organism; |
---|
77 | int numofref; |
---|
78 | Reference *reference; |
---|
79 | Comment comments; |
---|
80 | } GenBank; |
---|
81 | |
---|
82 | /* Macke data structure */ |
---|
83 | typedef struct { |
---|
84 | char *seqabbr; /* seq. abbrev. */ |
---|
85 | char *name; /* seq. full name */ |
---|
86 | int rna_or_dna; /* rna or dna */ |
---|
87 | char *atcc; /* CC# of seq. */ |
---|
88 | char *rna; /* Sequence methods, old version entry */ |
---|
89 | char *date; /* date of modification */ |
---|
90 | char *nbk; /* GenBank information -old version entry */ |
---|
91 | char *acs; /* accession number */ |
---|
92 | char *author; /* author of the first reference */ |
---|
93 | char *journal; /* journal of the first reference */ |
---|
94 | char *title; /* title of the first reference */ |
---|
95 | char *who; /* who key in the data */ |
---|
96 | char *strain; /* strain */ |
---|
97 | char *subspecies; /* subspecies */ |
---|
98 | int numofrem; /* num. of remarks */ |
---|
99 | char **remarks; /* remarks */ |
---|
100 | } Macke; |
---|
101 | |
---|
102 | /* Paup data structure */ |
---|
103 | typedef struct { |
---|
104 | int ntax; /* number of sequences */ |
---|
105 | int nchar; /* max number of chars per seq. */ |
---|
106 | int labelpos; /* Label start from left or right */ |
---|
107 | int missing; |
---|
108 | const char *equate; /* equal meaning char */ |
---|
109 | int interleave; /* interleave or sequential */ |
---|
110 | int datatype; /* rna in this case */ |
---|
111 | char gap; /* char of gap, default is '-' */ |
---|
112 | int gapmode; |
---|
113 | /* if sequence data is not too large, read into memory at once */ |
---|
114 | /* otherwise, read in one by one. */ |
---|
115 | } Paup; |
---|
116 | |
---|
117 | typedef struct { |
---|
118 | char *author; |
---|
119 | char *title; |
---|
120 | char *journal; |
---|
121 | char *processing; |
---|
122 | } Emblref; |
---|
123 | |
---|
124 | typedef struct { |
---|
125 | char *id; /* entry name */ |
---|
126 | char *dateu; /* date of last updated */ |
---|
127 | char *datec; /* date of created */ |
---|
128 | char *description; /* description line (DE) */ |
---|
129 | char *os; /* Organism species */ |
---|
130 | char *accession; /* accession number(s) */ |
---|
131 | char *keywords; /* keyword */ |
---|
132 | int numofref; /* num. of reference */ |
---|
133 | Emblref *reference; |
---|
134 | char *dr; /* database cross-reference */ |
---|
135 | Comment comments; /* comments */ |
---|
136 | } Embl; |
---|
137 | |
---|
138 | typedef struct { |
---|
139 | char *id; |
---|
140 | char *filename; |
---|
141 | int format; /* could be NBRF, UWGCG, RMBL, STADEN */ |
---|
142 | char defgap; |
---|
143 | int num_of_sequence; /* num of sequence after merging with gaps */ |
---|
144 | char *sequence; /* sequence data after mergin with gaps */ |
---|
145 | } Alma; |
---|
146 | |
---|
147 | typedef struct { |
---|
148 | char *id; /* locus */ |
---|
149 | char *description; |
---|
150 | } Nbrf; |
---|
151 | |
---|
152 | /* one sequence entry */ |
---|
153 | struct global_data { |
---|
154 | int numofseq; /* number of sequences */ |
---|
155 | int seq_length; /* sequence length */ |
---|
156 | int max; |
---|
157 | char *sequence; /* sequence data */ |
---|
158 | /* to read all the sequences into memory at one time */ |
---|
159 | char **ids; /* array of ids. */ |
---|
160 | char **seqs; /* array of sequence data */ |
---|
161 | int *lengths; /* array of sequence lengths */ |
---|
162 | int allocated; /* for how many sequences space has been allocated */ |
---|
163 | /* PAUP, PHYLIP, GCG, and PRINTABLE */ |
---|
164 | GenBank gbk; /* one GenBank entry */ |
---|
165 | Macke macke; /* one Macke entry */ |
---|
166 | Paup paup; /* one Paup entry */ |
---|
167 | Embl embl; |
---|
168 | Alma alma; |
---|
169 | Nbrf nbrf; |
---|
170 | }; |
---|
171 | |
---|
172 | extern struct global_data data; |
---|
173 | |
---|