1 | #define MAXARGS 100 |
---|
2 | |
---|
3 | typedef struct { |
---|
4 | char *str; |
---|
5 | sint *flag; |
---|
6 | int type; |
---|
7 | char **arg; |
---|
8 | } cmd_line_data; |
---|
9 | |
---|
10 | /* |
---|
11 | command line switches |
---|
12 | */ |
---|
13 | sint setoptions = -1; |
---|
14 | sint sethelp = -1; |
---|
15 | sint setinteractive = -1; |
---|
16 | sint setbatch = -1; |
---|
17 | sint setgapopen = -1; |
---|
18 | sint setgapext = -1; |
---|
19 | sint setpwgapopen = -1; |
---|
20 | sint setpwgapext = -1; |
---|
21 | sint setoutorder = -1; |
---|
22 | sint setbootlabels = -1; |
---|
23 | sint setpwmatrix = -1; |
---|
24 | sint setmatrix = -1; |
---|
25 | sint setpwdnamatrix = -1; |
---|
26 | sint setdnamatrix = -1; |
---|
27 | sint setnegative = -1; |
---|
28 | sint setnoweights = -1; |
---|
29 | sint setoutput = -1; |
---|
30 | sint setoutputtree = -1; |
---|
31 | sint setquicktree = -1; |
---|
32 | sint settype = -1; |
---|
33 | sint setcase = -1; |
---|
34 | sint setseqno = -1; |
---|
35 | |
---|
36 | sint setseqno_range = -1; |
---|
37 | sint setrange = -1; |
---|
38 | |
---|
39 | sint settransweight = -1; |
---|
40 | sint setseed = -1; |
---|
41 | sint setscore = -1; |
---|
42 | sint setwindow = -1; |
---|
43 | sint setktuple = -1; |
---|
44 | sint setkimura = -1; |
---|
45 | sint settopdiags = -1; |
---|
46 | sint setpairgap = -1; |
---|
47 | sint settossgaps = -1; |
---|
48 | sint setnopgap = -1; |
---|
49 | sint setnohgap = -1; |
---|
50 | sint setnovgap = -1; |
---|
51 | sint sethgapres = -1; |
---|
52 | sint setvgapres = -1; |
---|
53 | sint setuseendgaps = -1; |
---|
54 | sint setmaxdiv = -1; |
---|
55 | sint setgapdist = -1; |
---|
56 | sint setdebug = -1; |
---|
57 | sint setoutfile = -1; |
---|
58 | sint setinfile = -1; |
---|
59 | sint setprofile1 = -1; |
---|
60 | sint setprofile2 = -1; |
---|
61 | sint setalign = -1; |
---|
62 | sint setconvert = -1; |
---|
63 | sint setnewtree = -1; |
---|
64 | sint setusetree = -1; |
---|
65 | sint setnewtree1 = -1; |
---|
66 | sint setusetree1 = -1; |
---|
67 | sint setnewtree2 = -1; |
---|
68 | sint setusetree2 = -1; |
---|
69 | sint setbootstrap = -1; |
---|
70 | sint settree = -1; |
---|
71 | sint setprofile = -1; |
---|
72 | sint setsequences = -1; |
---|
73 | sint setsecstr1 = -1; |
---|
74 | sint setsecstr2 = -1; |
---|
75 | sint setsecstroutput = -1; |
---|
76 | sint sethelixgap = -1; |
---|
77 | sint setstrandgap = -1; |
---|
78 | sint setloopgap = -1; |
---|
79 | sint setterminalgap = -1; |
---|
80 | sint sethelixendin = -1; |
---|
81 | sint sethelixendout = -1; |
---|
82 | sint setstrandendin = -1; |
---|
83 | sint setstrandendout = -1; |
---|
84 | |
---|
85 | /* |
---|
86 | multiple alignment parameters |
---|
87 | */ |
---|
88 | float dna_gap_open = 15.0, dna_gap_extend = 6.66; |
---|
89 | float prot_gap_open = 10.0, prot_gap_extend = 0.2; |
---|
90 | sint profile_type = PROFILE; |
---|
91 | sint gap_dist = 4; |
---|
92 | sint output_order = ALIGNED; |
---|
93 | sint divergence_cutoff = 30; |
---|
94 | sint matnum = 3; |
---|
95 | char mtrxname[FILENAMELEN+1] = "gonnet"; |
---|
96 | sint dnamatnum = 1; |
---|
97 | char dnamtrxname[FILENAMELEN+1] = "iub"; |
---|
98 | char hyd_residues[] = "GPSNDQEKR"; |
---|
99 | Boolean no_weights = FALSE; |
---|
100 | Boolean neg_matrix = FALSE; |
---|
101 | Boolean no_hyd_penalties = FALSE; |
---|
102 | Boolean no_var_penalties = TRUE; |
---|
103 | Boolean no_pref_penalties = FALSE; |
---|
104 | Boolean use_endgaps = FALSE; |
---|
105 | Boolean endgappenalties = FALSE; |
---|
106 | Boolean reset_alignments_new = FALSE; /* DES */ |
---|
107 | Boolean reset_alignments_all = FALSE; /* DES */ |
---|
108 | sint output_struct_penalties = 0; |
---|
109 | sint struct_penalties1 = NONE; |
---|
110 | sint struct_penalties2 = NONE; |
---|
111 | Boolean use_ss1 = TRUE; |
---|
112 | Boolean use_ss2 = TRUE; |
---|
113 | sint helix_penalty = 4; |
---|
114 | sint strand_penalty = 4; |
---|
115 | sint loop_penalty = 1; |
---|
116 | sint helix_end_minus = 3; |
---|
117 | sint helix_end_plus = 0; |
---|
118 | sint strand_end_minus = 1; |
---|
119 | sint strand_end_plus = 1; |
---|
120 | sint helix_end_penalty = 2; |
---|
121 | sint strand_end_penalty = 2; |
---|
122 | Boolean use_ambiguities = FALSE; |
---|
123 | |
---|
124 | /* |
---|
125 | pairwise alignment parameters |
---|
126 | */ |
---|
127 | float dna_pw_go_penalty = 15.0, dna_pw_ge_penalty = 6.66; |
---|
128 | float prot_pw_go_penalty = 10.0, prot_pw_ge_penalty = 0.1; |
---|
129 | sint pw_matnum = 3; |
---|
130 | char pw_mtrxname[FILENAMELEN+1] = "gonnet"; |
---|
131 | sint pw_dnamatnum = 1; |
---|
132 | char pw_dnamtrxname[FILENAMELEN+1] = "iub"; |
---|
133 | char usermtrxname[FILENAMELEN+1], pw_usermtrxname[FILENAMELEN+1]; |
---|
134 | char dnausermtrxname[FILENAMELEN+1], pw_dnausermtrxname[FILENAMELEN+1]; |
---|
135 | |
---|
136 | Boolean quick_pairalign = FALSE; |
---|
137 | float transition_weight = 0.5; |
---|
138 | sint new_seq; |
---|
139 | |
---|
140 | /* |
---|
141 | quick pairwise alignment parameters |
---|
142 | */ |
---|
143 | sint dna_ktup = 2; /* default parameters for DNA */ |
---|
144 | sint dna_wind_gap = 5; |
---|
145 | sint dna_signif = 4; |
---|
146 | sint dna_window = 4; |
---|
147 | |
---|
148 | sint prot_ktup = 1; /* default parameters for proteins */ |
---|
149 | sint prot_wind_gap = 3; |
---|
150 | sint prot_signif = 5; |
---|
151 | sint prot_window = 5; |
---|
152 | Boolean percent=TRUE; |
---|
153 | Boolean tossgaps = FALSE; |
---|
154 | Boolean kimura = FALSE; |
---|
155 | |
---|
156 | |
---|
157 | sint boot_ntrials = 1000; |
---|
158 | unsigned sint boot_ran_seed = 111; |
---|
159 | |
---|
160 | |
---|
161 | sint debug = 0; |
---|
162 | |
---|
163 | Boolean explicit_dnaflag = FALSE; /* Explicit setting of sequence type on comm.line*/ |
---|
164 | Boolean lowercase = TRUE; /* Flag for GDE output - set on comm. line*/ |
---|
165 | Boolean cl_seq_numbers = FALSE; |
---|
166 | |
---|
167 | Boolean seqRange = FALSE; /* Ramu */ |
---|
168 | |
---|
169 | Boolean output_clustal = TRUE; |
---|
170 | Boolean output_gcg = FALSE; |
---|
171 | Boolean output_phylip = FALSE; |
---|
172 | Boolean output_nbrf = FALSE; |
---|
173 | Boolean output_gde = FALSE; |
---|
174 | Boolean output_nexus = FALSE; |
---|
175 | Boolean output_fasta = FALSE; |
---|
176 | |
---|
177 | Boolean showaln = TRUE; |
---|
178 | Boolean save_parameters = FALSE; |
---|
179 | |
---|
180 | /* DES */ |
---|
181 | Boolean output_tree_clustal = FALSE; |
---|
182 | Boolean output_tree_phylip = TRUE; |
---|
183 | Boolean output_tree_distances = FALSE; |
---|
184 | Boolean output_tree_nexus = FALSE; |
---|
185 | Boolean output_pim = FALSE; |
---|
186 | |
---|
187 | |
---|
188 | sint bootstrap_format = BS_BRANCH_LABELS; |
---|
189 | |
---|
190 | /*These are all the positively scoring groups that occur in the Gonnet Pam250 |
---|
191 | matrix. There are strong and weak groups, defined as strong score >0.5 and |
---|
192 | weak score =<0.5. Strong matching columns to be assigned ':' and weak matches |
---|
193 | assigned '.' in the clustal output format. |
---|
194 | */ |
---|
195 | |
---|
196 | char *res_cat1[] = { |
---|
197 | "STA", |
---|
198 | "NEQK", |
---|
199 | "NHQK", |
---|
200 | "NDEQ", |
---|
201 | "QHRK", |
---|
202 | "MILV", |
---|
203 | "MILF", |
---|
204 | "HY", |
---|
205 | "FYW", |
---|
206 | NULL }; |
---|
207 | |
---|
208 | char *res_cat2[] = { |
---|
209 | "CSA", |
---|
210 | "ATV", |
---|
211 | "SAG", |
---|
212 | "STNK", |
---|
213 | "STPA", |
---|
214 | "SGND", |
---|
215 | "SNDEQK", |
---|
216 | "NDEQHK", |
---|
217 | "NEQHRK", |
---|
218 | "FVLIM", |
---|
219 | "HFY", |
---|
220 | NULL }; |
---|
221 | |
---|
222 | |
---|
223 | |
---|
224 | static char *type_arg[] = { |
---|
225 | "protein", |
---|
226 | "dna", |
---|
227 | ""}; |
---|
228 | |
---|
229 | static char *bootlabels_arg[] = { |
---|
230 | "node", |
---|
231 | "branch", |
---|
232 | ""}; |
---|
233 | |
---|
234 | static char *outorder_arg[] = { |
---|
235 | "input", |
---|
236 | "aligned", |
---|
237 | ""}; |
---|
238 | |
---|
239 | static char *case_arg[] = { |
---|
240 | "lower", |
---|
241 | "upper", |
---|
242 | ""}; |
---|
243 | |
---|
244 | static char *seqno_arg[] = { |
---|
245 | "off", |
---|
246 | "on", |
---|
247 | ""}; |
---|
248 | |
---|
249 | static char *seqno_range_arg[] = { |
---|
250 | "off", |
---|
251 | "on", |
---|
252 | ""}; |
---|
253 | |
---|
254 | static char *score_arg[] = { |
---|
255 | "percent", |
---|
256 | "absolute", |
---|
257 | ""}; |
---|
258 | |
---|
259 | static char *output_arg[] = { |
---|
260 | "gcg", |
---|
261 | "gde", |
---|
262 | "pir", |
---|
263 | "phylip", |
---|
264 | "nexus", |
---|
265 | "fasta", |
---|
266 | ""}; |
---|
267 | |
---|
268 | static char *outputtree_arg[] = { |
---|
269 | "nj", |
---|
270 | "phylip", |
---|
271 | "dist", |
---|
272 | "nexus", |
---|
273 | ""}; |
---|
274 | |
---|
275 | static char *outputsecstr_arg[] = { |
---|
276 | "structure", |
---|
277 | "mask", |
---|
278 | "both", |
---|
279 | "none", |
---|
280 | ""}; |
---|
281 | |
---|
282 | /* |
---|
283 | command line initialisation |
---|
284 | |
---|
285 | type = 0 no argument |
---|
286 | type = 1 integer argument |
---|
287 | type = 2 float argument |
---|
288 | type = 3 string argument |
---|
289 | type = 4 filename |
---|
290 | type = 5 opts |
---|
291 | */ |
---|
292 | #define NOARG 0 |
---|
293 | #define INTARG 1 |
---|
294 | #define FLTARG 2 |
---|
295 | #define STRARG 3 |
---|
296 | #define FILARG 4 |
---|
297 | #define OPTARG 5 |
---|
298 | |
---|
299 | |
---|
300 | /* command line switches for DATA **************************/ |
---|
301 | cmd_line_data cmd_line_file[] = { |
---|
302 | "infile", &setinfile, FILARG, NULL, |
---|
303 | "profile1", &setprofile1, FILARG, NULL, |
---|
304 | "profile2", &setprofile2, FILARG, NULL, |
---|
305 | "", NULL, -1}; |
---|
306 | /* command line switches for VERBS **************************/ |
---|
307 | cmd_line_data cmd_line_verb[] = { |
---|
308 | "help", &sethelp, NOARG, NULL, |
---|
309 | "check", &sethelp, NOARG, NULL, |
---|
310 | "options", &setoptions, NOARG, NULL, |
---|
311 | "align", &setalign, NOARG, NULL, |
---|
312 | "newtree", &setnewtree, FILARG, NULL, |
---|
313 | "usetree", &setusetree, FILARG, NULL, |
---|
314 | "newtree1", &setnewtree1, FILARG, NULL, |
---|
315 | "usetree1", &setusetree1, FILARG, NULL, |
---|
316 | "newtree2", &setnewtree2, FILARG, NULL, |
---|
317 | "usetree2", &setusetree2, FILARG, NULL, |
---|
318 | "bootstrap", &setbootstrap, NOARG, NULL, |
---|
319 | "tree", &settree, NOARG, NULL, |
---|
320 | "quicktree", &setquicktree, NOARG, NULL, |
---|
321 | "convert", &setconvert, NOARG, NULL, |
---|
322 | "interactive", &setinteractive, NOARG, NULL, |
---|
323 | "batch", &setbatch, NOARG, NULL, |
---|
324 | "", NULL, -1}; |
---|
325 | /* command line switches for PARAMETERS **************************/ |
---|
326 | cmd_line_data cmd_line_para[] = { |
---|
327 | "type", &settype, OPTARG, type_arg, |
---|
328 | "profile", &setprofile, NOARG, NULL, |
---|
329 | "sequences", &setsequences, NOARG, NULL, |
---|
330 | "matrix", &setmatrix, FILARG, NULL, |
---|
331 | "dnamatrix", &setdnamatrix, FILARG, NULL, |
---|
332 | "negative", &setnegative, NOARG, NULL, |
---|
333 | "noweights", &setnoweights, NOARG, NULL, |
---|
334 | "gapopen", &setgapopen, FLTARG, NULL, |
---|
335 | "gapext", &setgapext, FLTARG, NULL, |
---|
336 | "endgaps", &setuseendgaps, NOARG, NULL, |
---|
337 | "nopgap", &setnopgap, NOARG, NULL, |
---|
338 | "nohgap", &setnohgap, NOARG, NULL, |
---|
339 | "novgap", &setnovgap, NOARG, NULL, |
---|
340 | "hgapresidues", &sethgapres, STRARG, NULL, |
---|
341 | "maxdiv", &setmaxdiv, INTARG, NULL, |
---|
342 | |
---|
343 | "gapdist", &setgapdist, INTARG, NULL, |
---|
344 | "pwmatrix", &setpwmatrix, FILARG, NULL, |
---|
345 | "pwdnamatrix", &setpwdnamatrix, FILARG, NULL, |
---|
346 | "pwgapopen", &setpwgapopen, FLTARG, NULL, |
---|
347 | "pwgapext", &setpwgapext, FLTARG, NULL, |
---|
348 | "ktuple", &setktuple, INTARG, NULL, |
---|
349 | "window", &setwindow, INTARG, NULL, |
---|
350 | "pairgap", &setpairgap, INTARG, NULL, |
---|
351 | "topdiags", &settopdiags, INTARG, NULL, |
---|
352 | "score", &setscore, OPTARG, score_arg, |
---|
353 | "transweight", &settransweight, FLTARG, NULL, |
---|
354 | "seed", &setseed, INTARG, NULL, |
---|
355 | "kimura", &setkimura, NOARG, NULL, |
---|
356 | "tossgaps", &settossgaps, NOARG, NULL, |
---|
357 | "bootlabels", &setbootlabels, OPTARG, bootlabels_arg, |
---|
358 | "debug", &setdebug, INTARG, NULL, |
---|
359 | "output", &setoutput, OPTARG, output_arg, |
---|
360 | "outputtree", &setoutputtree, OPTARG, outputtree_arg, |
---|
361 | "outfile", &setoutfile, FILARG, NULL, |
---|
362 | "outorder", &setoutorder, OPTARG, outorder_arg, |
---|
363 | "case", &setcase, OPTARG, case_arg, |
---|
364 | "seqnos", &setseqno, OPTARG, seqno_arg, |
---|
365 | |
---|
366 | "seqno_range", &setseqno_range, OPTARG, seqno_range_arg, /* this one should be on/off and */ |
---|
367 | "range", &setrange, STRARG, NULL, /* this one should be like 10:20 , messy option settings */ |
---|
368 | |
---|
369 | "nosecstr1", &setsecstr1, NOARG, NULL, |
---|
370 | "nosecstr2", &setsecstr2, NOARG, NULL, |
---|
371 | "secstrout", &setsecstroutput, OPTARG, outputsecstr_arg, |
---|
372 | "helixgap", &sethelixgap, INTARG, NULL, |
---|
373 | "strandgap", &setstrandgap, INTARG, NULL, |
---|
374 | "loopgap", &setloopgap, INTARG, NULL, |
---|
375 | "terminalgap", &setterminalgap, INTARG, NULL, |
---|
376 | "helixendin", &sethelixendin, INTARG, NULL, |
---|
377 | "helixendout", &sethelixendout, INTARG, NULL, |
---|
378 | "strandendin", &setstrandendin, INTARG, NULL, |
---|
379 | "strandendout",&setstrandendout, INTARG, NULL, |
---|
380 | |
---|
381 | "", NULL, -1}; |
---|
382 | |
---|
383 | |
---|