1 | /*#include "/us1/user/julie/dmalloc/malloc.h"*/ |
---|
2 | /*********************CLUSTALW.H*********************************************/ |
---|
3 | /****************************************************************************/ |
---|
4 | |
---|
5 | /* |
---|
6 | Main header file for ClustalW. Uncomment ONE of the following 4 lines |
---|
7 | depending on which compiler you wish to use. |
---|
8 | */ |
---|
9 | |
---|
10 | /*#define VMS 1 VAX or ALPHA VMS */ |
---|
11 | |
---|
12 | /*#define MAC 1 Think_C for Macintosh */ |
---|
13 | |
---|
14 | /*#define MSDOS 1 Turbo C for PC's */ |
---|
15 | |
---|
16 | #define UNIX 1 /*Ultrix/Decstation, Gnu C for |
---|
17 | Sun, IRIX/SGI, OSF1/ALPHA */ |
---|
18 | |
---|
19 | /***************************************************************************/ |
---|
20 | /***************************************************************************/ |
---|
21 | |
---|
22 | |
---|
23 | #include "general.h" |
---|
24 | |
---|
25 | #define MAXNAMES 30 /* Max chars read for seq. names */ |
---|
26 | #define MAXTITLES 60 /* Title length */ |
---|
27 | #define FILENAMELEN 256 /* Max. file name length */ |
---|
28 | |
---|
29 | #define UNKNOWN 0 |
---|
30 | #define EMBLSWISS 1 |
---|
31 | #define PIR 2 |
---|
32 | #define PEARSON 3 |
---|
33 | #define GDE 4 |
---|
34 | #define CLUSTAL 5 /* DES */ |
---|
35 | #define MSF 6 /* DES */ |
---|
36 | #define RSF 7 /* JULIE */ |
---|
37 | #define USER 8 /* DES */ |
---|
38 | #define PHYLIP 9 /* DES */ |
---|
39 | #define NEXUS 10/* DES */ |
---|
40 | #define FASTA 11/* Ramu */ |
---|
41 | |
---|
42 | #define NONE 0 |
---|
43 | #define SECST 1 |
---|
44 | #define GMASK 2 |
---|
45 | |
---|
46 | #define PROFILE 0 |
---|
47 | #define SEQUENCE 1 |
---|
48 | |
---|
49 | #define BS_NODE_LABELS 2 |
---|
50 | #define BS_BRANCH_LABELS 1 |
---|
51 | |
---|
52 | #define PAGE_LEN 22 /* Number of lines of help sent to screen */ |
---|
53 | |
---|
54 | #define PAGEWIDTH 80 /* maximum characters on output file page */ |
---|
55 | #define LINELENGTH 60 /* Output file line length */ |
---|
56 | #define GCG_LINELENGTH 50 |
---|
57 | |
---|
58 | #ifdef VMS /* Defaults for VAX VMS */ |
---|
59 | #define COMMANDSEP '/' |
---|
60 | #define DIRDELIM ']' /* Last character before file name in full file |
---|
61 | specs */ |
---|
62 | #define INT_SCALE_FACTOR 1000 /* Scaling factor to convert float to integer for profile scores */ |
---|
63 | |
---|
64 | #elif MAC |
---|
65 | #define COMMANDSEP '/' |
---|
66 | #define DIRDELIM ':' |
---|
67 | #define INT_SCALE_FACTOR 100 /* Scaling factor to convert float to integer for profile scores */ |
---|
68 | |
---|
69 | #elif MSDOS |
---|
70 | #define COMMANDSEP '/' |
---|
71 | #define DIRDELIM '\\' |
---|
72 | #define INT_SCALE_FACTOR 100 /* Scaling factor to convert float to integer for profile scores */ |
---|
73 | |
---|
74 | #elif UNIX |
---|
75 | #define COMMANDSEP '-' |
---|
76 | #define DIRDELIM '/' |
---|
77 | #define INT_SCALE_FACTOR 1000 /* Scaling factor to convert float to integer for profile scores */ |
---|
78 | #endif |
---|
79 | |
---|
80 | #define NUMRES 32 /* max size of comparison matrix */ |
---|
81 | |
---|
82 | #define INPUT 0 |
---|
83 | #define ALIGNED 1 |
---|
84 | |
---|
85 | #define LEFT 1 |
---|
86 | #define RIGHT 2 |
---|
87 | |
---|
88 | #define NODE 0 |
---|
89 | #define LEAF 1 |
---|
90 | |
---|
91 | #define GAPCOL 32 /* position of gap open penalty in profile */ |
---|
92 | #define LENCOL 33 /* position of gap extension penalty in profile */ |
---|
93 | |
---|
94 | typedef struct node { /* phylogenetic tree structure */ |
---|
95 | struct node *left; |
---|
96 | struct node *right; |
---|
97 | struct node *parent; |
---|
98 | float dist; |
---|
99 | sint leaf; |
---|
100 | int order; |
---|
101 | char name[64]; |
---|
102 | } stree, *treeptr; |
---|
103 | |
---|
104 | typedef struct { |
---|
105 | char title[30]; |
---|
106 | char string[30]; |
---|
107 | } MatMenuEntry; |
---|
108 | |
---|
109 | typedef struct { |
---|
110 | int noptions; |
---|
111 | MatMenuEntry opt[10]; |
---|
112 | } MatMenu; |
---|
113 | |
---|
114 | #define MAXMAT 10 |
---|
115 | |
---|
116 | typedef struct { |
---|
117 | int llimit; |
---|
118 | int ulimit; |
---|
119 | short *matptr; |
---|
120 | short *aa_xref; |
---|
121 | } SeriesMat; |
---|
122 | |
---|
123 | typedef struct { |
---|
124 | int nmat; |
---|
125 | SeriesMat mat[MAXMAT]; |
---|
126 | } UserMatSeries; |
---|
127 | |
---|
128 | |
---|
129 | /* |
---|
130 | Prototypes |
---|
131 | */ |
---|
132 | |
---|
133 | /* alnscore.c */ |
---|
134 | void aln_score(void); |
---|
135 | /* interface.c */ |
---|
136 | void parse_params(Boolean); |
---|
137 | void init_amenu(void); |
---|
138 | void init_interface(void); |
---|
139 | void main_menu(void); |
---|
140 | FILE *open_output_file(char *, char *, char *, char *); |
---|
141 | FILE *open_explicit_file(char *); |
---|
142 | sint seq_input(Boolean); |
---|
143 | Boolean open_alignment_output(char *); |
---|
144 | void create_alignment_output(sint fseq,sint lseq); |
---|
145 | void align(char *phylip_name); |
---|
146 | void profile_align(char *p1_tree_name,char *p2_tree_name);/* Align 2 alignments */ |
---|
147 | void make_tree(char *phylip_name); |
---|
148 | void get_tree(char *phylip_name); |
---|
149 | sint profile_input(void); /* read a profile */ |
---|
150 | void new_sequence_align(char *phylip_name); |
---|
151 | Boolean user_mat(char *, short *, short *); |
---|
152 | Boolean user_mat_series(char *, short *, short *); |
---|
153 | void get_help(char); |
---|
154 | void clustal_out(FILE *, sint, sint, sint, sint); |
---|
155 | void nbrf_out(FILE *, sint, sint, sint, sint); |
---|
156 | void gcg_out(FILE *, sint, sint, sint, sint); |
---|
157 | void phylip_out(FILE *, sint, sint, sint, sint); |
---|
158 | void gde_out(FILE *, sint, sint, sint, sint); |
---|
159 | void nexus_out(FILE *, sint, sint, sint, sint); |
---|
160 | void fasta_out(FILE *, sint, sint, sint, sint); |
---|
161 | void print_sec_struct_mask(int prf_length,char *mask,char *struct_mask); |
---|
162 | void fix_gaps(void); |
---|
163 | |
---|
164 | |
---|
165 | /* calcgapcoeff.c */ |
---|
166 | void calc_gap_coeff(char **alignment, sint *gaps, sint **profile, Boolean struct_penalties, |
---|
167 | char *gap_penalty_mask, sint first_seq, sint last_seq, |
---|
168 | sint prf_length, sint gapcoef, sint lencoef); |
---|
169 | /* calcprf1.c */ |
---|
170 | void calc_prf1(sint **profile, char **alignment, sint *gaps, sint matrix[NUMRES ][NUMRES ], |
---|
171 | sint *seq_weight, sint prf_length, sint first_seq, sint last_seq); |
---|
172 | /* calcprf2.c */ |
---|
173 | void calc_prf2(sint **profile, char **alignment, sint *seq_weight, sint prf_length, |
---|
174 | sint first_seq, sint last_seq); |
---|
175 | /* calctree.c */ |
---|
176 | void calc_seq_weights(sint first_seq, sint last_seq,sint *seq_weight); |
---|
177 | void create_sets(sint first_seq, sint last_seq); |
---|
178 | sint read_tree(char *treefile, sint first_seq, sint last_seq); |
---|
179 | void clear_tree(treeptr p); |
---|
180 | sint calc_similarities(sint nseqs); |
---|
181 | /* clustalw.c */ |
---|
182 | int main(int argc, char **argv); |
---|
183 | /* gcgcheck.c */ |
---|
184 | int SeqGCGCheckSum(char *seq, sint len); |
---|
185 | /* malign.c */ |
---|
186 | sint malign(sint istart,char *phylip_name); |
---|
187 | sint seqalign(sint istart,char *phylip_name); |
---|
188 | sint palign1(void); |
---|
189 | float countid(sint s1, sint s2); |
---|
190 | sint palign2(char *p1_tree_name,char *p2_tree_name); |
---|
191 | /* pairalign.c */ |
---|
192 | sint pairalign(sint istart, sint iend, sint jstart, sint jend); |
---|
193 | /* prfalign.c */ |
---|
194 | lint prfalign(sint *group, sint *aligned); |
---|
195 | /* random.c */ |
---|
196 | unsigned long linrand(unsigned long r); |
---|
197 | unsigned long addrand(unsigned long r); |
---|
198 | void addrandinit(unsigned long s); |
---|
199 | /* readmat.c */ |
---|
200 | void init_matrix(void); |
---|
201 | sint get_matrix(short *matptr, short *xref, sint matrix[NUMRES ][NUMRES ], Boolean neg_flag, |
---|
202 | sint scale); |
---|
203 | sint read_user_matrix(char *filename, short *usermat, short *xref); |
---|
204 | sint read_matrix_series(char *filename, short *usermat, short *xref); |
---|
205 | int getargs(char *inline1, char *args[], int max); |
---|
206 | /* sequence.c */ |
---|
207 | void fill_chartab(void); |
---|
208 | sint readseqs(sint first_seq); |
---|
209 | /* showpair.c */ |
---|
210 | void show_pair(sint istart, sint iend, sint jstart, sint jend); |
---|
211 | /* trees.c */ |
---|
212 | void phylogenetic_tree(char *phylip_name,char *clustal_name,char *dist_name, char *nexus_name, char *pim_name); |
---|
213 | void bootstrap_tree(char *phylip_name,char *clustal_name, char *nexus_name); |
---|
214 | sint dna_distance_matrix(FILE *tree); |
---|
215 | sint prot_distance_matrix(FILE *tree); |
---|
216 | void guide_tree(FILE *tree,int first_seq,sint nseqs); |
---|
217 | |
---|
218 | void calc_percidentity(FILE *pfile); |
---|
219 | |
---|
220 | /* util.c */ |
---|
221 | |
---|
222 | void alloc_aln(sint nseqs); |
---|
223 | void realloc_aln(sint first_seq,sint nseqs); |
---|
224 | void free_aln(sint nseqs); |
---|
225 | void alloc_seq(sint seq_no,sint length); |
---|
226 | void realloc_seq(sint seq_no,sint length); |
---|
227 | void free_seq(sint seq_no); |
---|
228 | |
---|
229 | void *ckalloc(size_t bytes); |
---|
230 | void *ckrealloc(void *ptr, size_t bytes); |
---|
231 | void *ckfree(void *ptr); |
---|
232 | char prompt_for_yes_no(char *title,char *prompt); |
---|
233 | void fatal(char *msg, ...); |
---|
234 | void error(char *msg, ...); |
---|
235 | void warning(char *msg, ...); |
---|
236 | void info(char *msg, ...); |
---|
237 | char *rtrim(char *str); |
---|
238 | char *blank_to_(char *str); |
---|
239 | char *upstr(char *str); |
---|
240 | char *lowstr(char *str); |
---|
241 | void getstr(char *instr, char *outstr); |
---|
242 | double getreal(char *instr, double minx, double maxx, double def); |
---|
243 | int getint(char *instr, int minx, int maxx, int def); |
---|
244 | void do_system(void); |
---|
245 | Boolean linetype(char *line, char *code); |
---|
246 | Boolean keyword(char *line, char *code); |
---|
247 | Boolean blankline(char *line); |
---|
248 | void get_path(char *str, char *path); |
---|
249 | |
---|
250 | |
---|