1 | #include "mltaln.h" |
---|
2 | #include "dp.h" |
---|
3 | |
---|
4 | #define MACHIGAI 0 |
---|
5 | #define OUTGAP0TRY 1 |
---|
6 | #define DEBUG 0 |
---|
7 | #define XXXXXXX 0 |
---|
8 | #define USE_PENALTY_EX 0 |
---|
9 | #define FASTMATCHCALC 1 |
---|
10 | |
---|
11 | |
---|
12 | static TLS float **impmtx = NULL; |
---|
13 | #if 1 // tditeration to naiveRscore_imp de tsukawareru. |
---|
14 | float imp_match_out_scR( int i1, int j1 ) |
---|
15 | { |
---|
16 | // fprintf( stderr, "imp+match = %f\n", impmtx[i1][j1] * fastathreshold ); |
---|
17 | // fprintf( stderr, "val = %f\n", impmtx[i1][j1] ); |
---|
18 | return( impmtx[i1][j1] ); |
---|
19 | } |
---|
20 | #endif |
---|
21 | |
---|
22 | static void imp_match_out_veadR( float *imp, int i1, int lgth2 ) |
---|
23 | { |
---|
24 | #if FASTMATCHCALC |
---|
25 | float *pt = impmtx[i1]; |
---|
26 | while( lgth2-- ) |
---|
27 | *imp++ += *pt++; |
---|
28 | #else |
---|
29 | int j; |
---|
30 | float *pt = impmtx[i1]; |
---|
31 | for( j=0; j<lgth2; j++ ) |
---|
32 | *imp++ += pt[j]; |
---|
33 | #endif |
---|
34 | } |
---|
35 | static void imp_match_out_vead_tateR( float *imp, int j1, int lgth1 ) |
---|
36 | { |
---|
37 | int i; |
---|
38 | for( i=0; i<lgth1; i++ ) |
---|
39 | *imp++ += impmtx[i][j1]; |
---|
40 | } |
---|
41 | |
---|
42 | #if 1 // tbfast.c kara yobareru. |
---|
43 | void imp_match_init_strictR( float *imp, int clus1, int clus2, int lgth1, int lgth2, char **seq1, char **seq2, double *eff1, double *eff2, LocalHom ***localhom, int forscore ) |
---|
44 | { |
---|
45 | int i, j, k1, k2, tmpint, start1, start2, end1, end2; |
---|
46 | static TLS int impalloclen = 0; |
---|
47 | float effij; |
---|
48 | double effijx; |
---|
49 | char *pt, *pt1, *pt2; |
---|
50 | static TLS char *nocount1 = NULL; |
---|
51 | static char *nocount2 = NULL; |
---|
52 | LocalHom *tmpptr; |
---|
53 | |
---|
54 | if( impalloclen < lgth1 + 2 || impalloclen < lgth2 + 2 ) |
---|
55 | { |
---|
56 | if( impmtx ) FreeFloatMtx( impmtx ); |
---|
57 | if( nocount1 ) free( nocount1 ); |
---|
58 | if( nocount2 ) free( nocount2 ); |
---|
59 | impalloclen = MAX( lgth1, lgth2 ) + 2; |
---|
60 | impmtx = AllocateFloatMtx( impalloclen, impalloclen ); |
---|
61 | nocount1 = AllocateCharVec( impalloclen ); |
---|
62 | nocount2 = AllocateCharVec( impalloclen ); |
---|
63 | } |
---|
64 | |
---|
65 | for( i=0; i<lgth1; i++ ) |
---|
66 | { |
---|
67 | for( j=0; j<clus1; j++ ) |
---|
68 | if( seq1[j][i] == '-' ) break; |
---|
69 | if( j != clus1 ) nocount1[i] = 1; |
---|
70 | else nocount1[i] = 0; |
---|
71 | } |
---|
72 | for( i=0; i<lgth2; i++ ) |
---|
73 | { |
---|
74 | for( j=0; j<clus2; j++ ) |
---|
75 | if( seq2[j][i] == '-' ) break; |
---|
76 | if( j != clus2 ) nocount2[i] = 1; |
---|
77 | else nocount2[i] = 0; |
---|
78 | } |
---|
79 | |
---|
80 | #if 0 |
---|
81 | fprintf( stderr, "nocount2 =\n" ); |
---|
82 | for( i = 0; i<impalloclen; i++ ) |
---|
83 | { |
---|
84 | fprintf( stderr, "nocount2[%d] = %d (%c)\n", i, nocount2[i], seq2[0][i] ); |
---|
85 | } |
---|
86 | #endif |
---|
87 | |
---|
88 | |
---|
89 | |
---|
90 | #if 0 |
---|
91 | fprintf( stderr, "eff1 in _init_strict = \n" ); |
---|
92 | for( i=0; i<clus1; i++ ) |
---|
93 | fprintf( stderr, "eff1[] = %f\n", eff1[i] ); |
---|
94 | for( i=0; i<clus2; i++ ) |
---|
95 | fprintf( stderr, "eff2[] = %f\n", eff2[i] ); |
---|
96 | #endif |
---|
97 | |
---|
98 | for( i=0; i<lgth1; i++ ) for( j=0; j<lgth2; j++ ) |
---|
99 | impmtx[i][j] = 0.0; |
---|
100 | effijx = fastathreshold; |
---|
101 | for( i=0; i<clus1; i++ ) |
---|
102 | { |
---|
103 | for( j=0; j<clus2; j++ ) |
---|
104 | { |
---|
105 | effij = (float)( eff1[i] * eff2[j] * effijx ); |
---|
106 | tmpptr = localhom[i][j]; |
---|
107 | while( tmpptr ) |
---|
108 | { |
---|
109 | // fprintf( stderr, "start1 = %d\n", tmpptr->start1 ); |
---|
110 | // fprintf( stderr, "end1 = %d\n", tmpptr->end1 ); |
---|
111 | // fprintf( stderr, "i = %d, seq1 = \n%s\n", i, seq1[i] ); |
---|
112 | // fprintf( stderr, "j = %d, seq2 = \n%s\n", j, seq2[j] ); |
---|
113 | pt = seq1[i]; |
---|
114 | tmpint = -1; |
---|
115 | while( *pt != 0 ) |
---|
116 | { |
---|
117 | if( *pt++ != '-' ) tmpint++; |
---|
118 | if( tmpint == tmpptr->start1 ) break; |
---|
119 | } |
---|
120 | start1 = pt - seq1[i] - 1; |
---|
121 | |
---|
122 | if( tmpptr->start1 == tmpptr->end1 ) end1 = start1; |
---|
123 | else |
---|
124 | { |
---|
125 | #if MACHIGAI |
---|
126 | while( *pt != 0 ) |
---|
127 | { |
---|
128 | // fprintf( stderr, "tmpint = %d, end1 = %d pos = %d\n", tmpint, tmpptr->end1, pt-seq1[i] ); |
---|
129 | if( tmpint == tmpptr->end1 ) break; |
---|
130 | if( *pt++ != '-' ) tmpint++; |
---|
131 | } |
---|
132 | end1 = pt - seq1[i] - 0; |
---|
133 | #else |
---|
134 | while( *pt != 0 ) |
---|
135 | { |
---|
136 | // fprintf( stderr, "tmpint = %d, end1 = %d pos = %d\n", tmpint, tmpptr->end1, pt-seq1[i] ); |
---|
137 | if( *pt++ != '-' ) tmpint++; |
---|
138 | if( tmpint == tmpptr->end1 ) break; |
---|
139 | } |
---|
140 | end1 = pt - seq1[i] - 1; |
---|
141 | #endif |
---|
142 | } |
---|
143 | |
---|
144 | pt = seq2[j]; |
---|
145 | tmpint = -1; |
---|
146 | while( *pt != 0 ) |
---|
147 | { |
---|
148 | if( *pt++ != '-' ) tmpint++; |
---|
149 | if( tmpint == tmpptr->start2 ) break; |
---|
150 | } |
---|
151 | start2 = pt - seq2[j] - 1; |
---|
152 | if( tmpptr->start2 == tmpptr->end2 ) end2 = start2; |
---|
153 | else |
---|
154 | { |
---|
155 | #if MACHIGAI |
---|
156 | while( *pt != 0 ) |
---|
157 | { |
---|
158 | if( tmpint == tmpptr->end2 ) break; |
---|
159 | if( *pt++ != '-' ) tmpint++; |
---|
160 | } |
---|
161 | end2 = pt - seq2[j] - 0; |
---|
162 | #else |
---|
163 | while( *pt != 0 ) |
---|
164 | { |
---|
165 | if( *pt++ != '-' ) tmpint++; |
---|
166 | if( tmpint == tmpptr->end2 ) break; |
---|
167 | } |
---|
168 | end2 = pt - seq2[j] - 1; |
---|
169 | #endif |
---|
170 | } |
---|
171 | // fprintf( stderr, "start1 = %d (%c), end1 = %d (%c), start2 = %d (%c), end2 = %d (%c)\n", start1, seq1[i][start1], end1, seq1[i][end1], start2, seq2[j][start2], end2, seq2[j][end2] ); |
---|
172 | // fprintf( stderr, "step 0\n" ); |
---|
173 | if( end1 - start1 != end2 - start2 ) |
---|
174 | { |
---|
175 | // fprintf( stderr, "CHUUI!!, start1 = %d, end1 = %d, start2 = %d, end2 = %d\n", start1, end1, start2, end2 ); |
---|
176 | } |
---|
177 | |
---|
178 | #if 1 |
---|
179 | k1 = start1; k2 = start2; |
---|
180 | pt1 = seq1[i] + k1; |
---|
181 | pt2 = seq2[j] + k2; |
---|
182 | while( *pt1 && *pt2 ) |
---|
183 | { |
---|
184 | if( *pt1 != '-' && *pt2 != '-' ) |
---|
185 | { |
---|
186 | // œÅ€ß€òÆóœÅ€Ë€«€±€Ê€€€è€Š€ËÃí°Õ€·€Æ²Œ€µ€€¡£ |
---|
187 | // impmtx[k1][k2] += tmpptr->wimportance * fastathreshold; |
---|
188 | // impmtx[k1][k2] += tmpptr->importance * effij; |
---|
189 | impmtx[k1][k2] += tmpptr->fimportance * effij; |
---|
190 | // fprintf( stderr, "#### impmtx[k1][k2] = %f, tmpptr->fimportance=%f, effij=%f\n", impmtx[k1][k2], tmpptr->fimportance, effij ); |
---|
191 | // fprintf( stderr, "mark, %d (%c) - %d (%c) \n", k1, *pt1, k2, *pt2 ); |
---|
192 | // fprintf( stderr, "%d (%c) - %d (%c) - %f\n", k1, *pt1, k2, *pt2, tmpptr->fimportance * effij ); |
---|
193 | k1++; k2++; |
---|
194 | pt1++; pt2++; |
---|
195 | } |
---|
196 | else if( *pt1 != '-' && *pt2 == '-' ) |
---|
197 | { |
---|
198 | // fprintf( stderr, "skip, %d (%c) - %d (%c) \n", k1, *pt1, k2, *pt2 ); |
---|
199 | k2++; pt2++; |
---|
200 | } |
---|
201 | else if( *pt1 == '-' && *pt2 != '-' ) |
---|
202 | { |
---|
203 | // fprintf( stderr, "skip, %d (%c) - %d (%c) \n", k1, *pt1, k2, *pt2 ); |
---|
204 | k1++; pt1++; |
---|
205 | } |
---|
206 | else if( *pt1 == '-' && *pt2 == '-' ) |
---|
207 | { |
---|
208 | // fprintf( stderr, "skip, %d (%c) - %d (%c) \n", k1, *pt1, k2, *pt2 ); |
---|
209 | k1++; pt1++; |
---|
210 | k2++; pt2++; |
---|
211 | } |
---|
212 | if( k1 > end1 || k2 > end2 ) break; |
---|
213 | } |
---|
214 | #else |
---|
215 | while( k1 <= end1 && k2 <= end2 ) |
---|
216 | { |
---|
217 | fprintf( stderr, "k1,k2=%d,%d - ", k1, k2 ); |
---|
218 | if( !nocount1[k1] && !nocount2[k2] ) |
---|
219 | { |
---|
220 | impmtx[k1][k2] += tmpptr->wimportance * eff1[i] * eff2[j] * fastathreshold; |
---|
221 | fprintf( stderr, "marked\n" ); |
---|
222 | } |
---|
223 | else |
---|
224 | fprintf( stderr, "no count\n" ); |
---|
225 | k1++; k2++; |
---|
226 | } |
---|
227 | #endif |
---|
228 | tmpptr = tmpptr->next; |
---|
229 | } |
---|
230 | } |
---|
231 | } |
---|
232 | #if 0 |
---|
233 | if( clus1 == 1 && clus2 == 6 ) |
---|
234 | { |
---|
235 | fprintf( stderr, "\n" ); |
---|
236 | fprintf( stderr, "seq1[0] = %s\n", seq1[0] ); |
---|
237 | fprintf( stderr, "seq2[0] = %s\n", seq2[0] ); |
---|
238 | fprintf( stderr, "impmtx = \n" ); |
---|
239 | for( k2=0; k2<lgth2; k2++ ) |
---|
240 | fprintf( stderr, "%6.3f ", (double)k2 ); |
---|
241 | fprintf( stderr, "\n" ); |
---|
242 | for( k1=0; k1<lgth1; k1++ ) |
---|
243 | { |
---|
244 | fprintf( stderr, "%d ", k1 ); |
---|
245 | for( k2=0; k2<3; k2++ ) |
---|
246 | fprintf( stderr, "%2.1f ", impmtx[k1][k2] ); |
---|
247 | fprintf( stderr, "\n" ); |
---|
248 | } |
---|
249 | exit( 1 ); |
---|
250 | } |
---|
251 | #endif |
---|
252 | } |
---|
253 | #endif |
---|
254 | |
---|
255 | |
---|
256 | static void match_calc( float *match, float **cpmx1, float **cpmx2, int i1, int lgth2, float **floatwork, int **intwork, int initialize ) |
---|
257 | { |
---|
258 | #if FASTMATCHCALC |
---|
259 | int j, l; |
---|
260 | float scarr[26]; |
---|
261 | float **cpmxpd = floatwork; |
---|
262 | int **cpmxpdn = intwork; |
---|
263 | float *matchpt, *cpmxpdpt, **cpmxpdptpt; |
---|
264 | int *cpmxpdnpt, **cpmxpdnptpt; |
---|
265 | if( initialize ) |
---|
266 | { |
---|
267 | int count = 0; |
---|
268 | for( j=0; j<lgth2; j++ ) |
---|
269 | { |
---|
270 | count = 0; |
---|
271 | for( l=0; l<26; l++ ) |
---|
272 | { |
---|
273 | if( cpmx2[l][j] ) |
---|
274 | { |
---|
275 | cpmxpd[j][count] = cpmx2[l][j]; |
---|
276 | cpmxpdn[j][count] = l; |
---|
277 | count++; |
---|
278 | } |
---|
279 | } |
---|
280 | cpmxpdn[j][count] = -1; |
---|
281 | } |
---|
282 | } |
---|
283 | |
---|
284 | { |
---|
285 | for( l=0; l<26; l++ ) |
---|
286 | { |
---|
287 | scarr[l] = 0.0; |
---|
288 | for( j=0; j<26; j++ ) |
---|
289 | scarr[l] += n_dis[j][l] * cpmx1[j][i1]; |
---|
290 | } |
---|
291 | matchpt = match; |
---|
292 | cpmxpdnptpt = cpmxpdn; |
---|
293 | cpmxpdptpt = cpmxpd; |
---|
294 | while( lgth2-- ) |
---|
295 | { |
---|
296 | *matchpt = 0.0; |
---|
297 | cpmxpdnpt = *cpmxpdnptpt++; |
---|
298 | cpmxpdpt = *cpmxpdptpt++; |
---|
299 | while( *cpmxpdnpt>-1 ) |
---|
300 | *matchpt += scarr[*cpmxpdnpt++] * *cpmxpdpt++; |
---|
301 | matchpt++; |
---|
302 | } |
---|
303 | } |
---|
304 | #else |
---|
305 | int j, k, l; |
---|
306 | float scarr[26]; |
---|
307 | float **cpmxpd = floatwork; |
---|
308 | int **cpmxpdn = intwork; |
---|
309 | // simple |
---|
310 | if( initialize ) |
---|
311 | { |
---|
312 | int count = 0; |
---|
313 | for( j=0; j<lgth2; j++ ) |
---|
314 | { |
---|
315 | count = 0; |
---|
316 | for( l=0; l<26; l++ ) |
---|
317 | { |
---|
318 | if( cpmx2[l][j] ) |
---|
319 | { |
---|
320 | cpmxpd[count][j] = cpmx2[l][j]; |
---|
321 | cpmxpdn[count][j] = l; |
---|
322 | count++; |
---|
323 | } |
---|
324 | } |
---|
325 | cpmxpdn[count][j] = -1; |
---|
326 | } |
---|
327 | } |
---|
328 | for( l=0; l<26; l++ ) |
---|
329 | { |
---|
330 | scarr[l] = 0.0; |
---|
331 | for( k=0; k<26; k++ ) |
---|
332 | scarr[l] += n_dis[k][l] * cpmx1[k][i1]; |
---|
333 | } |
---|
334 | for( j=0; j<lgth2; j++ ) |
---|
335 | { |
---|
336 | match[j] = 0.0; |
---|
337 | for( k=0; cpmxpdn[k][j]>-1; k++ ) |
---|
338 | match[j] += scarr[cpmxpdn[k][j]] * cpmxpd[k][j]; |
---|
339 | } |
---|
340 | #endif |
---|
341 | } |
---|
342 | |
---|
343 | static void Atracking_localhom( float *impwmpt, float *lasthorizontalw, float *lastverticalw, |
---|
344 | char **seq1, char **seq2, |
---|
345 | char **mseq1, char **mseq2, |
---|
346 | float **cpmx1, float **cpmx2, |
---|
347 | int **ijp, int icyc, int jcyc ) |
---|
348 | { |
---|
349 | int i, j, l, iin, jin, ifi, jfi, lgth1, lgth2, k; |
---|
350 | float wm; |
---|
351 | char *gaptable1, *gt1bk; |
---|
352 | char *gaptable2, *gt2bk; |
---|
353 | lgth1 = strlen( seq1[0] ); |
---|
354 | lgth2 = strlen( seq2[0] ); |
---|
355 | gt1bk = AllocateCharVec( lgth1+lgth2+1 ); |
---|
356 | gt2bk = AllocateCharVec( lgth1+lgth2+1 ); |
---|
357 | |
---|
358 | #if 0 |
---|
359 | for( i=0; i<lgth1; i++ ) |
---|
360 | { |
---|
361 | fprintf( stderr, "lastverticalw[%d] = %f\n", i, lastverticalw[i] ); |
---|
362 | } |
---|
363 | #endif |
---|
364 | |
---|
365 | if( outgap == 1 ) |
---|
366 | ; |
---|
367 | else |
---|
368 | { |
---|
369 | wm = lastverticalw[0]; |
---|
370 | for( i=0; i<lgth1; i++ ) |
---|
371 | { |
---|
372 | if( lastverticalw[i] >= wm ) |
---|
373 | { |
---|
374 | wm = lastverticalw[i]; |
---|
375 | iin = i; jin = lgth2-1; |
---|
376 | ijp[lgth1][lgth2] = +( lgth1 - i ); |
---|
377 | } |
---|
378 | } |
---|
379 | for( j=0; j<lgth2; j++ ) |
---|
380 | { |
---|
381 | if( lasthorizontalw[j] >= wm ) |
---|
382 | { |
---|
383 | wm = lasthorizontalw[j]; |
---|
384 | iin = lgth1-1; jin = j; |
---|
385 | ijp[lgth1][lgth2] = -( lgth2 - j ); |
---|
386 | } |
---|
387 | } |
---|
388 | } |
---|
389 | |
---|
390 | for( i=0; i<lgth1+1; i++ ) |
---|
391 | { |
---|
392 | ijp[i][0] = i + 1; |
---|
393 | } |
---|
394 | for( j=0; j<lgth2+1; j++ ) |
---|
395 | { |
---|
396 | ijp[0][j] = -( j + 1 ); |
---|
397 | } |
---|
398 | |
---|
399 | gaptable1 = gt1bk + lgth1+lgth2; |
---|
400 | *gaptable1 = 0; |
---|
401 | gaptable2 = gt2bk + lgth1+lgth2; |
---|
402 | *gaptable2 = 0; |
---|
403 | |
---|
404 | iin = lgth1; jin = lgth2; |
---|
405 | *impwmpt = 0.0; |
---|
406 | for( k=0; k<=lgth1+lgth2; k++ ) |
---|
407 | { |
---|
408 | if( ijp[iin][jin] < 0 ) |
---|
409 | { |
---|
410 | ifi = iin-1; jfi = jin+ijp[iin][jin]; |
---|
411 | } |
---|
412 | else if( ijp[iin][jin] > 0 ) |
---|
413 | { |
---|
414 | ifi = iin-ijp[iin][jin]; jfi = jin-1; |
---|
415 | } |
---|
416 | else |
---|
417 | { |
---|
418 | ifi = iin-1; jfi = jin-1; |
---|
419 | } |
---|
420 | l = iin - ifi; |
---|
421 | while( --l ) |
---|
422 | { |
---|
423 | *--gaptable1 = 'o'; |
---|
424 | *--gaptable2 = '-'; |
---|
425 | k++; |
---|
426 | } |
---|
427 | l= jin - jfi; |
---|
428 | while( --l ) |
---|
429 | { |
---|
430 | *--gaptable1 = '-'; |
---|
431 | *--gaptable2 = 'o'; |
---|
432 | k++; |
---|
433 | } |
---|
434 | if( iin == lgth1 || jin == lgth2 ) |
---|
435 | ; |
---|
436 | else |
---|
437 | { |
---|
438 | *impwmpt += imp_match_out_scR( iin, jin ); |
---|
439 | |
---|
440 | // fprintf( stderr, "impwm = %f (iin=%d, jin=%d) seq1=%c, seq2=%c\n", *impwmpt, iin, jin, seq1[0][iin], seq2[0][jin] ); |
---|
441 | } |
---|
442 | if( iin <= 0 || jin <= 0 ) break; |
---|
443 | *--gaptable1 = 'o'; |
---|
444 | *--gaptable2 = 'o'; |
---|
445 | k++; |
---|
446 | iin = ifi; jin = jfi; |
---|
447 | } |
---|
448 | |
---|
449 | for( i=0; i<icyc; i++ ) gapireru( mseq1[i], seq1[i], gaptable1 ); |
---|
450 | for( j=0; j<jcyc; j++ ) gapireru( mseq2[j], seq2[j], gaptable2 ); |
---|
451 | |
---|
452 | free( gt1bk ); |
---|
453 | free( gt2bk ); |
---|
454 | } |
---|
455 | |
---|
456 | static float Atracking( float *lasthorizontalw, float *lastverticalw, |
---|
457 | char **seq1, char **seq2, |
---|
458 | char **mseq1, char **mseq2, |
---|
459 | float **cpmx1, float **cpmx2, |
---|
460 | int **ijp, int icyc, int jcyc ) |
---|
461 | { |
---|
462 | int i, j, l, iin, jin, ifi, jfi, lgth1, lgth2, k; |
---|
463 | float wm; |
---|
464 | char *gaptable1, *gt1bk; |
---|
465 | char *gaptable2, *gt2bk; |
---|
466 | lgth1 = strlen( seq1[0] ); |
---|
467 | lgth2 = strlen( seq2[0] ); |
---|
468 | |
---|
469 | gt1bk = AllocateCharVec( lgth1+lgth2+1 ); |
---|
470 | gt2bk = AllocateCharVec( lgth1+lgth2+1 ); |
---|
471 | |
---|
472 | #if 0 |
---|
473 | for( i=0; i<lgth1; i++ ) |
---|
474 | { |
---|
475 | fprintf( stderr, "lastverticalw[%d] = %f\n", i, lastverticalw[i] ); |
---|
476 | } |
---|
477 | #endif |
---|
478 | |
---|
479 | if( outgap == 1 ) |
---|
480 | ; |
---|
481 | else |
---|
482 | { |
---|
483 | wm = lastverticalw[0]; |
---|
484 | for( i=0; i<lgth1; i++ ) |
---|
485 | { |
---|
486 | if( lastverticalw[i] >= wm ) |
---|
487 | { |
---|
488 | wm = lastverticalw[i]; |
---|
489 | iin = i; jin = lgth2-1; |
---|
490 | ijp[lgth1][lgth2] = +( lgth1 - i ); |
---|
491 | } |
---|
492 | } |
---|
493 | for( j=0; j<lgth2; j++ ) |
---|
494 | { |
---|
495 | if( lasthorizontalw[j] >= wm ) |
---|
496 | { |
---|
497 | wm = lasthorizontalw[j]; |
---|
498 | iin = lgth1-1; jin = j; |
---|
499 | ijp[lgth1][lgth2] = -( lgth2 - j ); |
---|
500 | } |
---|
501 | } |
---|
502 | } |
---|
503 | |
---|
504 | for( i=0; i<lgth1+1; i++ ) |
---|
505 | { |
---|
506 | ijp[i][0] = i + 1; |
---|
507 | } |
---|
508 | for( j=0; j<lgth2+1; j++ ) |
---|
509 | { |
---|
510 | ijp[0][j] = -( j + 1 ); |
---|
511 | } |
---|
512 | |
---|
513 | gaptable1 = gt1bk + lgth1+lgth2; |
---|
514 | *gaptable1 = 0; |
---|
515 | gaptable2 = gt2bk + lgth1+lgth2; |
---|
516 | *gaptable2 = 0; |
---|
517 | |
---|
518 | iin = lgth1; jin = lgth2; |
---|
519 | for( k=0; k<=lgth1+lgth2; k++ ) |
---|
520 | { |
---|
521 | if( ijp[iin][jin] < 0 ) |
---|
522 | { |
---|
523 | ifi = iin-1; jfi = jin+ijp[iin][jin]; |
---|
524 | } |
---|
525 | else if( ijp[iin][jin] > 0 ) |
---|
526 | { |
---|
527 | ifi = iin-ijp[iin][jin]; jfi = jin-1; |
---|
528 | } |
---|
529 | else |
---|
530 | { |
---|
531 | ifi = iin-1; jfi = jin-1; |
---|
532 | } |
---|
533 | l = iin - ifi; |
---|
534 | while( --l ) |
---|
535 | { |
---|
536 | *--gaptable1 = 'o'; |
---|
537 | *--gaptable2 = '-'; |
---|
538 | k++; |
---|
539 | } |
---|
540 | l= jin - jfi; |
---|
541 | while( --l ) |
---|
542 | { |
---|
543 | *--gaptable1 = '-'; |
---|
544 | *--gaptable2 = 'o'; |
---|
545 | k++; |
---|
546 | } |
---|
547 | if( iin <= 0 || jin <= 0 ) break; |
---|
548 | *--gaptable1 = 'o'; |
---|
549 | *--gaptable2 = 'o'; |
---|
550 | k++; |
---|
551 | iin = ifi; jin = jfi; |
---|
552 | } |
---|
553 | |
---|
554 | for( i=0; i<icyc; i++ ) gapireru( mseq1[i], seq1[i], gaptable1 ); |
---|
555 | for( j=0; j<jcyc; j++ ) gapireru( mseq2[j], seq2[j], gaptable2 ); |
---|
556 | |
---|
557 | free( gt1bk ); |
---|
558 | free( gt2bk ); |
---|
559 | |
---|
560 | return( 0.0 ); |
---|
561 | } |
---|
562 | |
---|
563 | float R__align( char **seq1, char **seq2, double *eff1, double *eff2, int icyc, int jcyc, int alloclen, LocalHom ***localhom, float *impmatch, char *sgap1, char *sgap2, char *egap1, char *egap2 ) |
---|
564 | /* score no keisan no sai motokaraaru gap no atukai ni mondai ga aru */ |
---|
565 | { |
---|
566 | // int k; |
---|
567 | register int i, j; |
---|
568 | int lasti, lastj; /* outgap == 0 -> lgth1, outgap == 1 -> lgth1+1 */ |
---|
569 | int lgth1, lgth2; |
---|
570 | int resultlen; |
---|
571 | float wm = 0.0; /* int ?????? */ |
---|
572 | float g; |
---|
573 | float *currentw, *previousw; |
---|
574 | // float fpenalty = (float)penalty; |
---|
575 | #if USE_PENALTY_EX |
---|
576 | float fpenalty_ex = (float)penalty_ex; |
---|
577 | #endif |
---|
578 | #if 1 |
---|
579 | float *wtmp; |
---|
580 | int *ijppt; |
---|
581 | float *mjpt, *prept, *curpt; |
---|
582 | int *mpjpt; |
---|
583 | #endif |
---|
584 | static TLS float mi, *m; |
---|
585 | static TLS int **ijp; |
---|
586 | static TLS int mpi, *mp; |
---|
587 | static TLS float *w1, *w2; |
---|
588 | static TLS float *match; |
---|
589 | static TLS float *initverticalw; /* kufuu sureba iranai */ |
---|
590 | static TLS float *lastverticalw; /* kufuu sureba iranai */ |
---|
591 | static TLS char **mseq1; |
---|
592 | static TLS char **mseq2; |
---|
593 | static TLS char **mseq; |
---|
594 | static TLS float *digf1; |
---|
595 | static TLS float *digf2; |
---|
596 | static TLS float *diaf1; |
---|
597 | static TLS float *diaf2; |
---|
598 | static TLS float *gapz1; |
---|
599 | static TLS float *gapz2; |
---|
600 | static TLS float *gapf1; |
---|
601 | static TLS float *gapf2; |
---|
602 | static TLS float *ogcp1g; |
---|
603 | static TLS float *ogcp2g; |
---|
604 | static TLS float *fgcp1g; |
---|
605 | static TLS float *fgcp2g; |
---|
606 | static TLS float *ogcp1; |
---|
607 | static TLS float *ogcp2; |
---|
608 | static TLS float *fgcp1; |
---|
609 | static TLS float *fgcp2; |
---|
610 | static TLS float **cpmx1; |
---|
611 | static TLS float **cpmx2; |
---|
612 | static TLS int **intwork; |
---|
613 | static TLS float **floatwork; |
---|
614 | static TLS int orlgth1 = 0, orlgth2 = 0; |
---|
615 | float fpenalty = (float)penalty; |
---|
616 | float tmppenal; |
---|
617 | float cumpenal; |
---|
618 | float *fgcp2pt; |
---|
619 | float *ogcp2pt; |
---|
620 | float fgcp1va; |
---|
621 | float ogcp1va; |
---|
622 | float kyokaipenal; |
---|
623 | |
---|
624 | |
---|
625 | |
---|
626 | #if 0 |
---|
627 | fprintf( stderr, "#### seq1[0] = %s\n", seq1[0] ); |
---|
628 | fprintf( stderr, "#### seq2[0] = %s\n", seq2[0] ); |
---|
629 | #endif |
---|
630 | if( orlgth1 == 0 ) |
---|
631 | { |
---|
632 | mseq1 = AllocateCharMtx( njob, 0 ); |
---|
633 | mseq2 = AllocateCharMtx( njob, 0 ); |
---|
634 | } |
---|
635 | |
---|
636 | |
---|
637 | lgth1 = strlen( seq1[0] ); |
---|
638 | lgth2 = strlen( seq2[0] ); |
---|
639 | #if 0 |
---|
640 | if( lgth1 == 0 || lgth2 == 0 ) |
---|
641 | { |
---|
642 | fprintf( stderr, "WARNING (Aalignmm): lgth1=%d, lgth2=%d\n", lgth1, lgth2 ); |
---|
643 | } |
---|
644 | #endif |
---|
645 | |
---|
646 | if( lgth1 > orlgth1 || lgth2 > orlgth2 ) |
---|
647 | { |
---|
648 | int ll1, ll2; |
---|
649 | |
---|
650 | if( orlgth1 > 0 && orlgth2 > 0 ) |
---|
651 | { |
---|
652 | FreeFloatVec( w1 ); |
---|
653 | FreeFloatVec( w2 ); |
---|
654 | FreeFloatVec( match ); |
---|
655 | FreeFloatVec( initverticalw ); |
---|
656 | FreeFloatVec( lastverticalw ); |
---|
657 | |
---|
658 | FreeFloatVec( m ); |
---|
659 | FreeIntVec( mp ); |
---|
660 | |
---|
661 | FreeCharMtx( mseq ); |
---|
662 | |
---|
663 | FreeFloatVec( digf1 ); |
---|
664 | FreeFloatVec( digf2 ); |
---|
665 | FreeFloatVec( diaf1 ); |
---|
666 | FreeFloatVec( diaf2 ); |
---|
667 | FreeFloatVec( gapz1 ); |
---|
668 | FreeFloatVec( gapz2 ); |
---|
669 | FreeFloatVec( gapf1 ); |
---|
670 | FreeFloatVec( gapf2 ); |
---|
671 | FreeFloatVec( ogcp1 ); |
---|
672 | FreeFloatVec( ogcp2 ); |
---|
673 | FreeFloatVec( fgcp1 ); |
---|
674 | FreeFloatVec( fgcp2 ); |
---|
675 | FreeFloatVec( ogcp1g ); |
---|
676 | FreeFloatVec( ogcp2g ); |
---|
677 | FreeFloatVec( fgcp1g ); |
---|
678 | FreeFloatVec( fgcp2g ); |
---|
679 | |
---|
680 | |
---|
681 | FreeFloatMtx( cpmx1 ); |
---|
682 | FreeFloatMtx( cpmx2 ); |
---|
683 | |
---|
684 | FreeFloatMtx( floatwork ); |
---|
685 | FreeIntMtx( intwork ); |
---|
686 | } |
---|
687 | |
---|
688 | ll1 = MAX( (int)(1.3*lgth1), orlgth1 ) + 100; |
---|
689 | ll2 = MAX( (int)(1.3*lgth2), orlgth2 ) + 100; |
---|
690 | |
---|
691 | #if DEBUG |
---|
692 | fprintf( stderr, "\ntrying to allocate (%d+%d)xn matrices ... ", ll1, ll2 ); |
---|
693 | #endif |
---|
694 | |
---|
695 | w1 = AllocateFloatVec( ll2+2 ); |
---|
696 | w2 = AllocateFloatVec( ll2+2 ); |
---|
697 | match = AllocateFloatVec( ll2+2 ); |
---|
698 | |
---|
699 | initverticalw = AllocateFloatVec( ll1+2 ); |
---|
700 | lastverticalw = AllocateFloatVec( ll1+2 ); |
---|
701 | |
---|
702 | m = AllocateFloatVec( ll2+2 ); |
---|
703 | mp = AllocateIntVec( ll2+2 ); |
---|
704 | |
---|
705 | mseq = AllocateCharMtx( njob, ll1+ll2 ); |
---|
706 | |
---|
707 | digf1 = AllocateFloatVec( ll1+2 ); |
---|
708 | digf2 = AllocateFloatVec( ll2+2 ); |
---|
709 | diaf1 = AllocateFloatVec( ll1+2 ); |
---|
710 | diaf2 = AllocateFloatVec( ll2+2 ); |
---|
711 | gapz1 = AllocateFloatVec( ll1+2 ); |
---|
712 | gapz2 = AllocateFloatVec( ll2+2 ); |
---|
713 | gapf1 = AllocateFloatVec( ll1+2 ); |
---|
714 | gapf2 = AllocateFloatVec( ll2+2 ); |
---|
715 | ogcp1 = AllocateFloatVec( ll1+2 ); |
---|
716 | ogcp2 = AllocateFloatVec( ll2+2 ); |
---|
717 | fgcp1 = AllocateFloatVec( ll1+2 ); |
---|
718 | fgcp2 = AllocateFloatVec( ll2+2 ); |
---|
719 | ogcp1g = AllocateFloatVec( ll1+2 ); |
---|
720 | ogcp2g = AllocateFloatVec( ll2+2 ); |
---|
721 | fgcp1g = AllocateFloatVec( ll1+2 ); |
---|
722 | fgcp2g = AllocateFloatVec( ll2+2 ); |
---|
723 | |
---|
724 | cpmx1 = AllocateFloatMtx( 26, ll1+2 ); |
---|
725 | cpmx2 = AllocateFloatMtx( 26, ll2+2 ); |
---|
726 | |
---|
727 | #if FASTMATCHCALC |
---|
728 | floatwork = AllocateFloatMtx( MAX( ll1, ll2 )+2, 26 ); |
---|
729 | intwork = AllocateIntMtx( MAX( ll1, ll2 )+2, 27 ); |
---|
730 | #else |
---|
731 | floatwork = AllocateFloatMtx( 26, MAX( ll1, ll2 )+2 ); |
---|
732 | intwork = AllocateIntMtx( 26, MAX( ll1, ll2 )+2 ); |
---|
733 | #endif |
---|
734 | |
---|
735 | #if DEBUG |
---|
736 | fprintf( stderr, "succeeded\n" ); |
---|
737 | #endif |
---|
738 | |
---|
739 | orlgth1 = ll1 - 100; |
---|
740 | orlgth2 = ll2 - 100; |
---|
741 | } |
---|
742 | |
---|
743 | |
---|
744 | for( i=0; i<icyc; i++ ) |
---|
745 | { |
---|
746 | mseq1[i] = mseq[i]; |
---|
747 | seq1[i][lgth1] = 0; |
---|
748 | } |
---|
749 | for( j=0; j<jcyc; j++ ) |
---|
750 | { |
---|
751 | mseq2[j] = mseq[icyc+j]; |
---|
752 | seq2[j][lgth2] = 0; |
---|
753 | } |
---|
754 | |
---|
755 | |
---|
756 | if( orlgth1 > commonAlloc1 || orlgth2 > commonAlloc2 ) |
---|
757 | { |
---|
758 | int ll1, ll2; |
---|
759 | |
---|
760 | if( commonAlloc1 && commonAlloc2 ) |
---|
761 | { |
---|
762 | FreeIntMtx( commonIP ); |
---|
763 | } |
---|
764 | |
---|
765 | ll1 = MAX( orlgth1, commonAlloc1 ); |
---|
766 | ll2 = MAX( orlgth2, commonAlloc2 ); |
---|
767 | |
---|
768 | #if DEBUG |
---|
769 | fprintf( stderr, "\n\ntrying to allocate %dx%d matrices ... ", ll1+1, ll2+1 ); |
---|
770 | #endif |
---|
771 | |
---|
772 | commonIP = AllocateIntMtx( ll1+10, ll2+10 ); |
---|
773 | |
---|
774 | #if DEBUG |
---|
775 | fprintf( stderr, "succeeded\n\n" ); |
---|
776 | #endif |
---|
777 | |
---|
778 | commonAlloc1 = ll1; |
---|
779 | commonAlloc2 = ll2; |
---|
780 | } |
---|
781 | ijp = commonIP; |
---|
782 | |
---|
783 | #if 0 |
---|
784 | { |
---|
785 | float t = 0.0; |
---|
786 | for( i=0; i<icyc; i++ ) |
---|
787 | t += eff1[i]; |
---|
788 | fprintf( stderr, "## totaleff = %f\n", t ); |
---|
789 | } |
---|
790 | #endif |
---|
791 | |
---|
792 | cpmx_calc_new( seq1, cpmx1, eff1, lgth1, icyc ); |
---|
793 | cpmx_calc_new( seq2, cpmx2, eff2, lgth2, jcyc ); |
---|
794 | |
---|
795 | if( sgap1 ) |
---|
796 | { |
---|
797 | new_OpeningGapCount_zure( ogcp1g, icyc, seq1, eff1, lgth1, sgap1, egap1 ); |
---|
798 | new_OpeningGapCount_zure( ogcp2g, jcyc, seq2, eff2, lgth2, sgap2, egap2 ); |
---|
799 | new_FinalGapCount_zure( fgcp1g, icyc, seq1, eff1, lgth1, sgap1, egap1 ); |
---|
800 | new_FinalGapCount_zure( fgcp2g, jcyc, seq2, eff2, lgth2, sgap2, egap2 ); |
---|
801 | getdigapfreq_part( digf1, icyc, seq1, eff1, lgth1, sgap1, egap1 ); |
---|
802 | getdigapfreq_part( digf2, jcyc, seq2, eff2, lgth2, sgap2, egap2 ); |
---|
803 | getdiaminofreq_part( diaf1, icyc, seq1, eff1, lgth1, sgap1, egap1 ); |
---|
804 | getdiaminofreq_part( diaf2, jcyc, seq2, eff2, lgth2, sgap2, egap2 ); |
---|
805 | getgapfreq( gapf1, icyc, seq1, eff1, lgth1 ); |
---|
806 | getgapfreq( gapf2, jcyc, seq2, eff2, lgth2 ); |
---|
807 | getgapfreq_zure_part( gapz1, icyc, seq1, eff1, lgth1, sgap1 ); |
---|
808 | getgapfreq_zure_part( gapz2, jcyc, seq2, eff2, lgth2, sgap1 ); |
---|
809 | } |
---|
810 | else |
---|
811 | { |
---|
812 | st_OpeningGapCount( ogcp1g, icyc, seq1, eff1, lgth1 ); |
---|
813 | st_OpeningGapCount( ogcp2g, jcyc, seq2, eff2, lgth2 ); |
---|
814 | st_FinalGapCount_zure( fgcp1g, icyc, seq1, eff1, lgth1 ); |
---|
815 | st_FinalGapCount_zure( fgcp2g, jcyc, seq2, eff2, lgth2 ); |
---|
816 | getdigapfreq_st( digf1, icyc, seq1, eff1, lgth1 ); |
---|
817 | getdigapfreq_st( digf2, jcyc, seq2, eff2, lgth2 ); |
---|
818 | getdiaminofreq_x( diaf1, icyc, seq1, eff1, lgth1 ); |
---|
819 | getdiaminofreq_x( diaf2, jcyc, seq2, eff2, lgth2 ); |
---|
820 | getgapfreq( gapf1, icyc, seq1, eff1, lgth1 ); |
---|
821 | getgapfreq( gapf2, jcyc, seq2, eff2, lgth2 ); |
---|
822 | getgapfreq_zure( gapz1, icyc, seq1, eff1, lgth1 ); |
---|
823 | getgapfreq_zure( gapz2, jcyc, seq2, eff2, lgth2 ); |
---|
824 | } |
---|
825 | |
---|
826 | #if 0 |
---|
827 | for( i=0; i<lgth1; i++ ) |
---|
828 | fprintf( stderr, "ogcp1[%d]=%f\n", i, ogcp1[i] ); |
---|
829 | #endif |
---|
830 | |
---|
831 | currentw = w1; |
---|
832 | previousw = w2; |
---|
833 | |
---|
834 | match_calc( initverticalw, cpmx2, cpmx1, 0, lgth1, floatwork, intwork, 1 ); |
---|
835 | if( localhom ) |
---|
836 | imp_match_out_vead_tateR( initverticalw, 0, lgth1 ); // 060306 |
---|
837 | |
---|
838 | match_calc( currentw, cpmx1, cpmx2, 0, lgth2, floatwork, intwork, 1 ); |
---|
839 | if( localhom ) |
---|
840 | imp_match_out_veadR( currentw, 0, lgth2 ); // 060306 |
---|
841 | |
---|
842 | |
---|
843 | #if 0 // -> tbfast.c |
---|
844 | if( localhom ) |
---|
845 | imp_match_calc( currentw, icyc, jcyc, lgth1, lgth2, seq1, seq2, eff1, eff2, localhom, 1, 0 ); |
---|
846 | |
---|
847 | #endif |
---|
848 | |
---|
849 | |
---|
850 | |
---|
851 | kyokaipenal = 0.0; |
---|
852 | if( outgap == 1 ) |
---|
853 | { |
---|
854 | g = 0.0; |
---|
855 | |
---|
856 | g += ogcp1g[0] * ( 1.0-ogcp2g[0] ) * fpenalty * 0.5; |
---|
857 | // if( g ) fprintf( stderr, "init-match penal1=%f, %c-%c\n", g, seq1[0][0], seq2[0][0] ); |
---|
858 | |
---|
859 | g += ogcp2g[0] * ( 1.0-ogcp1g[0] ) * fpenalty * 0.5; |
---|
860 | // if( g ) fprintf( stderr, "init-match penal2=%f, %c-%c\n", g, seq1[0][0], seq2[0][0] ); |
---|
861 | |
---|
862 | g += fgcp1g[0] * ( 1.0-fgcp2g[0] ) * fpenalty * 0.5; |
---|
863 | // if( g ) fprintf( stderr, "match penal1=%f, %c-%c\n", g, seq1[0][i], seq2[0][j] ); |
---|
864 | |
---|
865 | g += fgcp2g[0] * ( 1.0-fgcp1g[0] ) * fpenalty * 0.5; |
---|
866 | // if( g ) fprintf( stderr, "match penal2=%f, %c-%c\n", g, seq1[0][i], seq2[0][j] ); |
---|
867 | |
---|
868 | kyokaipenal = g; |
---|
869 | initverticalw[0] += g; |
---|
870 | currentw[0] += g; |
---|
871 | |
---|
872 | cumpenal = 0.0; |
---|
873 | for( i=1; i<lgth1+1; i++ ) |
---|
874 | { |
---|
875 | // initverticalw[i] += ( ogcp1[0] + fgcp1[i-1] ) ; |
---|
876 | |
---|
877 | // Q tmppenal = ( (1.0-gapz2[0])*(1.0-ogcp1g[0]+fgcp1g[0]-digf1[0]) ) * 0.5 * fpenalty; // mada |
---|
878 | // tmppenal = ( (1.0-gapf2[j-1])*(1.0-ogcp1g[i]+fgcp1g[i]) + gapf2[j-1]*(1.0-digf1[i]-diaf1[i]) ) * 0.5 * fpenalty; // mada |
---|
879 | tmppenal = ( (1.0-gapz2[0])*(1.0-ogcp1g[0]+0.0) + gapz2[0]*(1.0-digf1[0] - diaf1[0]) ) * 0.5 * fpenalty; // mada |
---|
880 | // tmppenal = ( (1.0-0.0)*(1.0-ogcp1g[0]+0.0) + 0.0*(1.0-0.0-0.0) ) * 0.5 * fpenalty; // mada |
---|
881 | // tmppenal = 0.5 * fpenalty; |
---|
882 | // tmppenal -= ( (1.0-0.0) * (1.0-diaf1[0]) + 0.0 ) * 0.5 * fpenalty; // 0. |
---|
883 | // tmppenal -= ( (1.0-gapf2[j-1]) * ogcp1g[i] + gapf2[j-1] ) * 0.5 * fpenalty; |
---|
884 | // fprintf( stderr, "0,0<-%d,%d, tmppenal 1 = %f\n", i, 0, tmppenal ); |
---|
885 | initverticalw[i] += tmppenal; |
---|
886 | |
---|
887 | // Q tmppenal = ( (1.0-gapz2[1])*(1.0-fgcp1g[i]+ogcp1g[i]-digf1[i]) ) * 0.5 * fpenalty; // mada |
---|
888 | // tmppenal = ( (1.0-gapf2[j])*(1.0-fgcp1g[i]+ogcp1g[i]) + gapf2[j]*(1.0-digf1[i]-diaf1[i]) ) * 0.5 * fpenalty; // mada |
---|
889 | tmppenal = ( (1.0-gapz2[1])*(1.0-fgcp1g[i]+ogcp1g[i]) + gapz2[1]*(1.0-digf1[i]-diaf1[i]) ) * 0.5 * fpenalty; // mada |
---|
890 | // tmppenal = ( (1.0-gapf2[0])*(1.0-fgcp1g[i]+ogcp1g[i]) + gapf2[0]*(1.0-digf1[i]-diaf1[i]) ) * 0.5 * fpenalty; // mada |
---|
891 | // tmppenal = 0.5 * fpenalty; |
---|
892 | // tmppenal -= ( (1.0-gapf2[0]) * (1.0-diaf1[i]) + gapf2[0] ) * 0.5 * fpenalty; |
---|
893 | // tmppenal -= ( (1.0-gapf2[j]) * fgcp1g[i] + gapf2[j] ) * 0.5 * fpenalty; |
---|
894 | initverticalw[i] += tmppenal; |
---|
895 | cumpenal += ( ogcp1g[i-0] + fgcp1g[i] ) * fpenalty * 0.5; |
---|
896 | initverticalw[i] += cumpenal - fgcp1g[i] * fpenalty * 0.5; // honto ha iru |
---|
897 | // fprintf( stderr, "0,0<-%d,%d, tmppenal 2 = %f, cumpenal=%f, fgcp1g[i]=%f, ogcp1g[i]=%f\n", i, 0, tmppenal, cumpenal, fgcp1g[i], ogcp1g[i] ); |
---|
898 | |
---|
899 | } |
---|
900 | cumpenal = 0.0; |
---|
901 | for( j=1; j<lgth2+1; j++ ) |
---|
902 | { |
---|
903 | // currentw[j] += ( ogcp2[0] + fgcp2[j-1] ) ; |
---|
904 | |
---|
905 | // tmppenal = ( (1.0-gapz1[0])*(1.0-ogcp2g[0]+fgcp2g[0]-digf2[0]) ) * 0.5 * fpenalty; // mada |
---|
906 | // tmppenal = ( (1.0-gapf1[i-1])*(1.0-ogcp2g[j]+fgcp2g[j]) + gapf1[i-1]*(1.0-digf2[j]-diaf2[j]) ) * 0.5 * fpenalty; // mada |
---|
907 | tmppenal = ( (1.0-gapz1[0])*(1.0-ogcp2g[0]+0.0) + gapz1[0]*(1.0-digf2[j]-diaf2[j]) ) * 0.5 * fpenalty; // mada |
---|
908 | // tmppenal = ( (1.0-0.0)*(1.0-ogcp2g[0]+0.0) + 0.0*(1.0-0.0-0.0) ) * 0.5 * fpenalty; // mada |
---|
909 | // tmppenal = 0.5 * fpenalty; |
---|
910 | // tmppenal -= ( (1.0-0.0) * (1.0-diaf2[0]) + 0.0 ) * 0.5 * fpenalty; // 0. |
---|
911 | // tmppenal -= ( (1.0-gapf1[0]) * fgcp2g[j] + gapf1[0] ) * 0.5 * fpenalty; |
---|
912 | // fprintf( stderr, "0,0<-%d,%d, tmppenal 3 = %f\n", 0, j, tmppenal ); |
---|
913 | currentw[j] += tmppenal; |
---|
914 | |
---|
915 | // tmppenal = ( (1.0-gapz1[1])*(1.0-fgcp2g[j]+ogcp2g[j]-digf2[j]) ) * 0.5 * fpenalty; // mada |
---|
916 | // tmppenal = ( (1.0-gapf1[i])*(1.0-fgcp2g[j]+ogcp2g[j]) + gapf1[i]*(1.0-digf2[j]-diaf2[j]) ) * 0.5 * fpenalty; // mada |
---|
917 | tmppenal = ( (1.0-gapz1[1])*(1.0-fgcp2g[j]+ogcp2g[j]) + gapz1[1]*(1.0-digf2[j]-diaf2[j]) ) * 0.5 * fpenalty; // mada |
---|
918 | // tmppenal = ( (1.0-gapf1[0])*(1.0-fgcp2g[j]+ogcp2g[j]) + gapf1[0]*(1.0-digf2[j]-diaf2[j]) ) * 0.5 * fpenalty; // mada |
---|
919 | // tmppenal = 0.5 * fpenalty; |
---|
920 | // tmppenal -= ( (1.0-gapf1[0]) * (1.0-diaf2[j]) + gapf1[0] ) * 0.5 * fpenalty; |
---|
921 | // tmppenal -= ( (1.0-gapf1[0]) * ogcp2g[j] + gapf1[i-1] ) * 0.5 * fpenalty; |
---|
922 | // fprintf( stderr, "0,0<-%d,%d, tmppenal 4 = %f\n", 0, j, tmppenal ); |
---|
923 | currentw[j] += tmppenal; |
---|
924 | cumpenal += ( ogcp2g[j-0] + fgcp2g[j] ) * fpenalty * 0.5; |
---|
925 | currentw[j] += cumpenal - fgcp2g[j] * fpenalty * 0.5; // honto ha iru |
---|
926 | } |
---|
927 | } |
---|
928 | #if OUTGAP0TRY |
---|
929 | else |
---|
930 | { |
---|
931 | for( j=1; j<lgth2+1; j++ ) |
---|
932 | currentw[j] -= offset * j / 2.0; |
---|
933 | for( i=1; i<lgth1+1; i++ ) |
---|
934 | initverticalw[i] -= offset * i / 2.0; |
---|
935 | } |
---|
936 | #endif |
---|
937 | |
---|
938 | m[0] = 0.0; // iranai |
---|
939 | for( j=1; j<lgth2+1; ++j ) |
---|
940 | { |
---|
941 | mp[j] = 0; |
---|
942 | m[j] = currentw[j-1] + 10000 * fpenalty; //iinoka? |
---|
943 | } |
---|
944 | if( lgth2 == 0 ) |
---|
945 | lastverticalw[0] = 0.0; // Falign kara yobaretatoki kounarukanousei ari |
---|
946 | else |
---|
947 | lastverticalw[0] = currentw[lgth2-1]; |
---|
948 | |
---|
949 | if( outgap ) lasti = lgth1+1; else lasti = lgth1; |
---|
950 | |
---|
951 | #if XXXXXXX |
---|
952 | fprintf( stderr, "currentw = \n" ); |
---|
953 | for( i=0; i<lgth1+1; i++ ) |
---|
954 | { |
---|
955 | fprintf( stderr, "%5.2f ", currentw[i] ); |
---|
956 | } |
---|
957 | fprintf( stderr, "\n" ); |
---|
958 | fprintf( stderr, "initverticalw = \n" ); |
---|
959 | for( i=0; i<lgth2+1; i++ ) |
---|
960 | { |
---|
961 | fprintf( stderr, "%5.2f ", initverticalw[i] ); |
---|
962 | } |
---|
963 | fprintf( stderr, "\n" ); |
---|
964 | fprintf( stderr, "fcgp\n" ); |
---|
965 | for( i=0; i<lgth1; i++ ) |
---|
966 | fprintf( stderr, "fgcp1[%d]=%f\n", i, ogcp1[i] ); |
---|
967 | for( i=0; i<lgth2; i++ ) |
---|
968 | fprintf( stderr, "fgcp2[%d]=%f\n", i, ogcp2[i] ); |
---|
969 | #endif |
---|
970 | |
---|
971 | for( i=1; i<lasti; i++ ) |
---|
972 | { |
---|
973 | wtmp = previousw; |
---|
974 | previousw = currentw; |
---|
975 | currentw = wtmp; |
---|
976 | |
---|
977 | previousw[0] = initverticalw[i-1]; |
---|
978 | |
---|
979 | match_calc( currentw, cpmx1, cpmx2, i, lgth2, floatwork, intwork, 0 ); |
---|
980 | #if XXXXXXX |
---|
981 | fprintf( stderr, "\n" ); |
---|
982 | fprintf( stderr, "i=%d\n", i ); |
---|
983 | fprintf( stderr, "currentw = \n" ); |
---|
984 | for( j=0; j<lgth2; j++ ) |
---|
985 | { |
---|
986 | fprintf( stderr, "%5.2f ", currentw[j] ); |
---|
987 | } |
---|
988 | fprintf( stderr, "\n" ); |
---|
989 | #endif |
---|
990 | if( localhom ) |
---|
991 | { |
---|
992 | // fprintf( stderr, "Calling imp_match_calc (o) lgth = %d, i = %d\n", lgth1, i ); |
---|
993 | #if 0 |
---|
994 | imp_match_out_veadR( currentw, i, lgth2 ); |
---|
995 | #else |
---|
996 | imp_match_out_veadR( currentw, i, lgth2 ); |
---|
997 | #endif |
---|
998 | } |
---|
999 | #if XXXXXXX |
---|
1000 | fprintf( stderr, "\n" ); |
---|
1001 | fprintf( stderr, "i=%d\n", i ); |
---|
1002 | fprintf( stderr, "currentw = \n" ); |
---|
1003 | for( j=0; j<lgth2; j++ ) |
---|
1004 | { |
---|
1005 | fprintf( stderr, "%5.2f ", currentw[j] ); |
---|
1006 | } |
---|
1007 | fprintf( stderr, "\n" ); |
---|
1008 | #endif |
---|
1009 | currentw[0] = initverticalw[i]; |
---|
1010 | |
---|
1011 | mpi = 0; |
---|
1012 | mi = previousw[0] + 10000 * fpenalty; |
---|
1013 | |
---|
1014 | ijppt = ijp[i] + 1; |
---|
1015 | mjpt = m + 1; |
---|
1016 | prept = previousw; |
---|
1017 | curpt = currentw + 1; |
---|
1018 | mpjpt = mp + 1; |
---|
1019 | fgcp2pt = fgcp2; |
---|
1020 | ogcp2pt = ogcp2 + 1; |
---|
1021 | fgcp1va = fgcp1[i-1]; |
---|
1022 | ogcp1va = ogcp1[i]; |
---|
1023 | lastj = lgth2+1; |
---|
1024 | for( j=1; j<lastj; j++ ) |
---|
1025 | { |
---|
1026 | wm = *prept; |
---|
1027 | |
---|
1028 | g = ogcp1g[i] * ( 1.0-ogcp2g[j] ) * fpenalty * 0.5; |
---|
1029 | // if( g && i==j ) fprintf( stderr, "match penal1=%f, %c-%c\n", g, seq1[0][i], seq2[0][j] ); |
---|
1030 | wm += g; |
---|
1031 | |
---|
1032 | g = ogcp2g[j] * ( 1.0-ogcp1g[i] ) * fpenalty * 0.5; |
---|
1033 | // if( g && i==j ) fprintf( stderr, "match penal2=%f, %c-%c\n", g, seq1[0][i], seq2[0][j] ); |
---|
1034 | wm += g; |
---|
1035 | |
---|
1036 | g = fgcp1g[i] * ( 1.0-fgcp2g[j] ) * fpenalty * 0.5; |
---|
1037 | // if( g && i==j ) fprintf( stderr, "match penal3=%f, %c-%c\n", g, seq1[0][i], seq2[0][j] ); |
---|
1038 | wm += g; |
---|
1039 | |
---|
1040 | g = fgcp2g[j] * ( 1.0-fgcp1g[i] ) * fpenalty * 0.5; |
---|
1041 | // if( g && i==j ) fprintf( stderr, "match penal4=%f, %c-%c\n", g, seq1[0][i], seq2[0][j] ); |
---|
1042 | wm += g; |
---|
1043 | |
---|
1044 | #if 0 |
---|
1045 | g = fgcp2g[j] * ogcp1g[i] * fpenalty * 0.5; |
---|
1046 | if( g ) fprintf( stderr, "match m-penal5=%f, %c-%c\n", g, seq1[0][i], seq2[0][j] ); |
---|
1047 | wm -= g; |
---|
1048 | |
---|
1049 | g = fgcp1g[i] * ogcp2g[j] * fpenalty * 0.5; |
---|
1050 | if( g ) fprintf( stderr, "match m-penal6=%f, %c-%c\n", g, seq1[0][i], seq2[0][j] ); |
---|
1051 | wm -= g; |
---|
1052 | #endif |
---|
1053 | |
---|
1054 | *ijppt = 0; |
---|
1055 | |
---|
1056 | #if 0 |
---|
1057 | fprintf( stderr, "%5.0f->", wm ); |
---|
1058 | #endif |
---|
1059 | #if 0 |
---|
1060 | fprintf( stderr, "%5.0f?", g ); |
---|
1061 | #endif |
---|
1062 | // tmppenal = ( (1.0-gapz1[i+1])*(1.0-fgcp2g[j]+ogcp2g[j]-digf2[j]) ) * 0.5 * fpenalty; // mada |
---|
1063 | // tmppenal = ( (1.0-gapf1[i])*(1.0-fgcp2g[j]) + gapf1[i]*(1.0-digf2[j]-diaf2[j]) ) * 0.5 * fpenalty; // mada |
---|
1064 | tmppenal = ( (1.0-gapz1[i+1])*(1.0-fgcp2g[j]+ogcp2g[j]) + gapz1[i+1]*(1.0-digf2[j]-diaf2[j]) ) * 0.5 * fpenalty; // mada |
---|
1065 | // tmppenal = ( (1.0-gapf1[i])*(1.0-fgcp2g[j]+ogcp2g[j]) + gapf1[i]*(1.0-digf2[j]-diaf2[j]) ) * 0.5 * fpenalty; // mada |
---|
1066 | // tmppenal = 0.5 * fpenalty; |
---|
1067 | // tmppenal -= ( (1.0-gapf1[i])*(1.0-diaf2[j]) + gapf1[i] ) * 0.5 * fpenalty; |
---|
1068 | // tmppenal -= ( (1.0-gapf1[i])*fgcp2g[j] + gapf1[i] ) * 0.5 * fpenalty; |
---|
1069 | // tmppenal = *fgcp2pt-fpenalty*0.5*gapf1[i]; |
---|
1070 | // tmppenal = *fgcp2pt; |
---|
1071 | if( (g=mi+tmppenal) > wm ) |
---|
1072 | { |
---|
1073 | // fprintf( stderr, "jump i start=%f (j=%d, fgcp2g[j]=%f, digf2[j]=%f, diaf2[j]=%f), %c-%c\n", g-mi, j, fgcp2g[j], digf2[j], diaf2[j], seq1[0][i], seq2[0][j] ); |
---|
1074 | wm = g; |
---|
1075 | *ijppt = -( j - mpi ); |
---|
1076 | } |
---|
1077 | // tmppenal = ( (1.0-gapz1[i])*(1.0-ogcp2g[j]+fgcp2g[j]-digf2[j]) ) * 0.5 * fpenalty; // mada |
---|
1078 | tmppenal = ( (1.0-gapz1[i])*(1.0-ogcp2g[j]+fgcp2g[j]) + gapz1[i]*(1.0-digf2[j]-diaf2[j]) ) * 0.5 * fpenalty; // mada |
---|
1079 | // tmppenal = ( (1.0-gapf1[i-1])*(1.0-ogcp2g[j]+fgcp2g[j]) + gapf1[i-1]*(1.0-digf2[j]-diaf2[j]) ) * 0.5 * fpenalty; // mada |
---|
1080 | // tmppenal = *ogcp2pt-fpenalty*0.5*(gapf2[j-1]+gapf1[i-1]); |
---|
1081 | // tmppenal = 0.5 * fpenalty; |
---|
1082 | // tmppenal -= ( (1.0-gapf1[i-1])*(1.0-diaf2[j]) + gapf1[i-1] ) * 0.5 * fpenalty; |
---|
1083 | // tmppenal -= ( (1.0-gapf1[i-1])*ogcp2g[j] + gapf1[i-1] ) * 0.5 * fpenalty; |
---|
1084 | // tmppenal = *prept+*ogcp2pt; |
---|
1085 | if( (g=*prept+tmppenal) >= mi ) |
---|
1086 | { |
---|
1087 | // fprintf( stderr, "jump i end=%f, %c-%c\n", g-*prept, seq1[0][i-1], seq2[0][j-1] ); |
---|
1088 | mi = g; |
---|
1089 | mpi = j-1; |
---|
1090 | } |
---|
1091 | else if( j != 1 ) |
---|
1092 | { |
---|
1093 | mi += ( ogcp2g[j-0] + fgcp2g[j] ) * fpenalty * 0.5; // CHUUI honto ha iru |
---|
1094 | // fprintf( stderr, "%c%c/%c%c exp, og=%f,fg=%f\n", '=', '=', seq2[0][j-1], seq2[0][j], ogcp2g[j-0] * fpenalty*0.5, fgcp2g[j] * fpenalty*0.5 ); |
---|
1095 | } |
---|
1096 | #if USE_PENALTY_EX |
---|
1097 | mi += fpenalty_ex; |
---|
1098 | #endif |
---|
1099 | |
---|
1100 | #if 0 |
---|
1101 | fprintf( stderr, "%5.0f?", g ); |
---|
1102 | #endif |
---|
1103 | // tmppenal = ( (1.0-gapz2[j+1])*(1.0-fgcp1g[i]+ogcp1g[i]-digf1[i]) ) * 0.5 * fpenalty; // mada |
---|
1104 | tmppenal = ( (1.0-gapz2[j+1])*(1.0-fgcp1g[i]+ogcp1g[i]) + gapz2[j+1]*(1.0-digf1[i]-diaf1[i]) ) * 0.5 * fpenalty; // mada |
---|
1105 | // tmppenal = ( (1.0-gapf2[j])*(1.0-fgcp1g[i]+ogcp1g[i]) + gapf2[j]*(1.0-digf1[i]-diaf1[i]) ) * 0.5 * fpenalty; // mada |
---|
1106 | // tmppenal = ( (1.0-gapf2[j])*(1.0-fgcp1g[i]) + gapf2[j]*(1.0-digf1[i]-diaf1[i]) ) * 0.5 * fpenalty; // mada |
---|
1107 | // tmppenal = 0.5 * fpenalty; |
---|
1108 | // tmppenal -= ( (1.0-gapf2[j])*(1.0-diaf1[i]) + gapf2[j] ) * 0.5 * fpenalty; |
---|
1109 | // tmppenal -= ( (1.0-gapf2[j])*fgcp1g[i] + gapf2[j] ) * 0.5 * fpenalty; |
---|
1110 | // tmppenal = fgcp1va-fpenalty*0.5*gapf2[j]; |
---|
1111 | // tmppenal = fgcp1va; |
---|
1112 | if( (g=*mjpt+tmppenal) > wm ) |
---|
1113 | { |
---|
1114 | // if( seq1[0][i] == 'Y' && seq2[0][j] == 'B' ) |
---|
1115 | // fprintf( stderr, "jump j start=%f, %c-%c\n", g-*mjpt, seq1[0][i], seq2[0][j] ); |
---|
1116 | wm = g; |
---|
1117 | *ijppt = +( i - *mpjpt ); |
---|
1118 | } |
---|
1119 | // tmppenal = ( (1.0-gapz2[j])*(1.0-ogcp1g[i]+fgcp1g[i]-digf1[i]) ) * 0.5 * fpenalty; // mada |
---|
1120 | tmppenal = ( (1.0-gapz2[j])*(1.0-ogcp1g[i]+fgcp1g[i]) + gapz2[j]*(1.0-digf1[i]-diaf1[i]) ) * 0.5 * fpenalty; // mada |
---|
1121 | // tmppenal = ( (1.0-gapf2[j-1])*(1.0-ogcp1g[i]+fgcp1g[i]) + gapf2[j-1]*(1.0-digf1[i]-diaf1[i]) ) * 0.5 * fpenalty; // mada |
---|
1122 | // tmppenal = ( (1.0-gapf2[j-1])*(1.0-ogcp1g[i]) + gapf2[j-1]*(1.0-digf1[i]-diaf1[i]) ) * 0.5 * fpenalty; // mada |
---|
1123 | // tmppenal = ogcp1va-fpenalty*0.5*(gapf1[i-1]+gapf2[j-1]); |
---|
1124 | // tmppenal = 0.5 * fpenalty; |
---|
1125 | // tmppenal -= ( (1.0-gapf2[j-1]) * (1.0-diaf1[i]) + gapf2[j-1] ) * 0.5 * fpenalty; |
---|
1126 | // tmppenal -= ( (1.0-gapf2[j-1]) * ogcp1g[i] + gapf2[j-1] ) * 0.5 * fpenalty; |
---|
1127 | // tmppenal = 0.5 * fpenalty - ( (1.0-gapf2[j-1]) * (ogcp1g[i]) + gapf2[j-1] ) * ( 0.5 * fpenalty ); |
---|
1128 | // tmppenal = ogcp1va-fpenalty*0.5*gapf2[j-1]; |
---|
1129 | // tmppenal = ogcp1va; |
---|
1130 | if( (g=*prept+tmppenal) >= *mjpt ) |
---|
1131 | { |
---|
1132 | // if( seq1[0][i] == 'T' && seq2[0][j] == 'T' ) |
---|
1133 | // fprintf( stderr, "jump j end=%f, %c-%c\n", g-*prept, seq1[0][i-1], seq2[0][j-1] ); |
---|
1134 | *mjpt = g; |
---|
1135 | *mpjpt = i-1; |
---|
1136 | } |
---|
1137 | else if( i != 1 ) |
---|
1138 | { |
---|
1139 | m[j] += ( ogcp1g[i-0] + fgcp1g[i] ) * fpenalty * 0.5; // CHUUI honto ha iru |
---|
1140 | // fprintf( stderr, "%c%c/%c%c exp, og=%f,fg=%f\n", seq1[0][i-1], seq1[0][i], '=', '=', ogcp1g[i-0] * fpenalty*0.5, fgcp1g[i] * fpenalty*0.5 ); |
---|
1141 | } |
---|
1142 | #if USE_PENALTY_EX |
---|
1143 | m[j] += fpenalty_ex; |
---|
1144 | #endif |
---|
1145 | |
---|
1146 | #if 0 |
---|
1147 | fprintf( stderr, "%5.0f ", wm ); |
---|
1148 | #endif |
---|
1149 | *curpt++ += wm; |
---|
1150 | ijppt++; |
---|
1151 | mjpt++; |
---|
1152 | prept++; |
---|
1153 | mpjpt++; |
---|
1154 | fgcp2pt++; |
---|
1155 | ogcp2pt++; |
---|
1156 | } |
---|
1157 | lastverticalw[i] = currentw[lgth2-1]; |
---|
1158 | } |
---|
1159 | |
---|
1160 | // fprintf( stderr, "wm = %f\n", wm ); |
---|
1161 | |
---|
1162 | #if OUTGAP0TRY |
---|
1163 | if( !outgap ) |
---|
1164 | { |
---|
1165 | for( j=1; j<lgth2+1; j++ ) |
---|
1166 | currentw[j] -= offset * ( lgth2 - j ) / 2.0; |
---|
1167 | for( i=1; i<lgth1+1; i++ ) |
---|
1168 | lastverticalw[i] -= offset * ( lgth1 - i / 2.0); |
---|
1169 | } |
---|
1170 | #endif |
---|
1171 | |
---|
1172 | /* |
---|
1173 | fprintf( stderr, "\n" ); |
---|
1174 | for( i=0; i<icyc; i++ ) fprintf( stderr,"%s\n", seq1[i] ); |
---|
1175 | fprintf( stderr, "#####\n" ); |
---|
1176 | for( j=0; j<jcyc; j++ ) fprintf( stderr,"%s\n", seq2[j] ); |
---|
1177 | fprintf( stderr, "====>" ); |
---|
1178 | for( i=0; i<icyc; i++ ) strcpy( mseq1[i], seq1[i] ); |
---|
1179 | for( j=0; j<jcyc; j++ ) strcpy( mseq2[j], seq2[j] ); |
---|
1180 | */ |
---|
1181 | if( localhom ) |
---|
1182 | { |
---|
1183 | Atracking_localhom( impmatch, currentw, lastverticalw, seq1, seq2, mseq1, mseq2, cpmx1, cpmx2, ijp, icyc, jcyc ); |
---|
1184 | } |
---|
1185 | else |
---|
1186 | Atracking( currentw, lastverticalw, seq1, seq2, mseq1, mseq2, cpmx1, cpmx2, ijp, icyc, jcyc ); |
---|
1187 | |
---|
1188 | // fprintf( stderr, "### impmatch = %f\n", *impmatch ); |
---|
1189 | |
---|
1190 | resultlen = strlen( mseq1[0] ); |
---|
1191 | if( alloclen < resultlen || resultlen > N ) |
---|
1192 | { |
---|
1193 | fprintf( stderr, "alloclen=%d, resultlen=%d, N=%d\n", alloclen, resultlen, N ); |
---|
1194 | ErrorExit( "LENGTH OVER!\n" ); |
---|
1195 | } |
---|
1196 | |
---|
1197 | |
---|
1198 | for( i=0; i<icyc; i++ ) strcpy( seq1[i], mseq1[i] ); |
---|
1199 | for( j=0; j<jcyc; j++ ) strcpy( seq2[j], mseq2[j] ); |
---|
1200 | /* |
---|
1201 | fprintf( stderr, "\n" ); |
---|
1202 | for( i=0; i<icyc; i++ ) fprintf( stderr, "%s\n", mseq1[i] ); |
---|
1203 | fprintf( stderr, "#####\n" ); |
---|
1204 | for( j=0; j<jcyc; j++ ) fprintf( stderr, "%s\n", mseq2[j] ); |
---|
1205 | */ |
---|
1206 | |
---|
1207 | fprintf( stderr, "wm = %f\n", wm ); |
---|
1208 | |
---|
1209 | |
---|
1210 | return( wm ); |
---|
1211 | } |
---|
1212 | |
---|
1213 | |
---|