void foldrna( int nseq1, int nseq2, char **seq1, char **seq2, double *eff1, double *eff2, RNApair ***grouprna1, RNApair ***grouprna2, float **impmtx, int *gapmap1, int *gapmap2, RNApair *additionalpair ) { int i, j; // int ui, uj; // int uiup, ujup; int uido, ujdo; static TLS char **useq1, **useq2; static TLS char **oseq1, **oseq2, **oseq1r, **oseq2r, *odir1, *odir2; static TLS RNApair **pairprob1, **pairprob2; static TLS RNApair *pairpt1, *pairpt2; int lgth1 = strlen( seq1[0] ); int lgth2 = strlen( seq2[0] ); static TLS float **impmtx2; static TLS float **map; // double lenfac; float prob; int **sgapmap1, **sgapmap2; char *nogapdum; float **tbppmtx; // fprintf( stderr, "nseq1=%d, lgth1=%d\n", nseq1, lgth1 ); useq1 = AllocateCharMtx( nseq1, lgth1+10 ); useq2 = AllocateCharMtx( nseq2, lgth2+10 ); oseq1 = AllocateCharMtx( nseq1, lgth1+10 ); oseq2 = AllocateCharMtx( nseq2, lgth2+10 ); oseq1r = AllocateCharMtx( nseq1, lgth1+10 ); oseq2r = AllocateCharMtx( nseq2, lgth2+10 ); odir1 = AllocateCharVec( lgth1+10 ); odir2 = AllocateCharVec( lgth2+10 ); sgapmap1 = AllocateIntMtx( nseq1, lgth1+1 ); sgapmap2 = AllocateIntMtx( nseq2, lgth2+1 ); nogapdum = AllocateCharVec( MAX( lgth1, lgth2 ) ); pairprob1 = (RNApair **)calloc( lgth1, sizeof( RNApair *) ); pairprob2 = (RNApair **)calloc( lgth2, sizeof( RNApair *) ); map = AllocateFloatMtx( lgth1, lgth2 ); impmtx2 = AllocateFloatMtx( lgth1, lgth2 ); tbppmtx = AllocateFloatMtx( lgth1, lgth2 ); for( i=0; i<nseq1; i++ ) strcpy( useq1[i], seq1[i] ); for( i=0; i<nseq2; i++ ) strcpy( useq2[i], seq2[i] ); for( i=0; i<nseq1; i++ ) strcpy( oseq1[i], seq1[i] ); for( i=0; i<nseq2; i++ ) strcpy( oseq2[i], seq2[i] ); for( i=0; i<nseq1; i++ ) commongappick_record( 1, useq1+i, sgapmap1[i] ); for( i=0; i<nseq2; i++ ) commongappick_record( 1, useq2+i, sgapmap2[i] ); for( i=0; i<lgth1; i++ ) { pairprob1[i] = (RNApair *)calloc( 1, sizeof( RNApair ) ); pairprob1[i][0].bestpos = -1; pairprob1[i][0].bestscore = -1; } for( i=0; i<lgth2; i++ ) { pairprob2[i] = (RNApair *)calloc( 1, sizeof( RNApair ) ); pairprob2[i][0].bestpos = -1; pairprob2[i][0].bestscore = -1; } utot( nseq1, lgth1, oseq1 ); utot( nseq2, lgth2, oseq2 ); // fprintf( stderr, "folding group1\n" ); // rnalocal( oseq1, useq1, eff1, eff1, nseq1, nseq1, lgth1+10, pair1 ); /* base-pairing probability of group 1 */ if( rnaprediction == 'r' ) rnaalifoldcall( oseq1, nseq1, pairprob1 ); else mccaskillextract( oseq1, useq1, nseq1, pairprob1, grouprna1, sgapmap1, eff1 ); // fprintf( stderr, "folding group2\n" ); // rnalocal( oseq2, useq2, eff2, eff2, nseq2, nseq2, lgth2+10, pair2 ); /* base-pairing probability of group 2 */ if( rnaprediction == 'r' ) rnaalifoldcall( oseq2, nseq2, pairprob2 ); else mccaskillextract( oseq2, useq2, nseq2, pairprob2, grouprna2, sgapmap2, eff2 ); #if 0 makerseq( oseq1, oseq1r, odir1, pairprob1, nseq1, lgth1 ); makerseq( oseq2, oseq2r, odir2, pairprob2, nseq2, lgth2 ); fprintf( stderr, "%s\n", odir2 ); for( i=0; i<nseq1; i++ ) { fprintf( stdout, ">ori\n%s\n", oseq1[0] ); fprintf( stdout, ">rev\n%s\n", oseq1r[0] ); } #endif /* similarity score */ Lalignmm_hmout( oseq1, oseq2, eff1, eff2, nseq1, nseq2, 10000, NULL, NULL, NULL, NULL, map ); if( 1 ) { if( RNAscoremtx == 'n' ) { for( i=0; i<lgth1; i++ ) for( j=0; j<lgth2; j++ ) { // impmtx2[i][j] = osoiaveragescore( nseq1, nseq2, oseq1, oseq2, eff1, eff2, i, j ) * consweight_multi; impmtx2[i][j] = 0.0; } } else if( RNAscoremtx == 'r' ) { for( i=0; i<lgth1; i++ ) for( j=0; j<lgth2; j++ ) { tbppmtx[i][j] = 1.0; impmtx2[i][j] = 0.0; } for( i=0; i<lgth1; i++ ) for( pairpt1=pairprob1[i]; pairpt1->bestpos!=-1; pairpt1++ ) { for( j=0; j<lgth2; j++ ) for( pairpt2=pairprob2[j]; pairpt2->bestpos!=-1; pairpt2++ ) { uido = pairpt1->bestpos; ujdo = pairpt2->bestpos; prob = pairpt1->bestscore * pairpt2->bestscore; if( uido > -1 && ujdo > -1 ) { if( uido > i && j > ujdo ) { impmtx2[i][j] += prob * pairedribosumscore53( nseq1, nseq2, oseq1, oseq2, eff1, eff2, i, j, uido, ujdo ) * consweight_multi; tbppmtx[i][j] -= prob; } else if( i < uido && j < ujdo ) { impmtx2[i][j] += prob * pairedribosumscore35( nseq1, nseq2, oseq1, oseq2, eff1, eff2, i, j, uido, ujdo ) * consweight_multi; tbppmtx[i][j] -= prob; } } } } for( i=0; i<lgth1; i++ ) { for( j=0; j<lgth2; j++ ) { impmtx2[i][j] += tbppmtx[i][j] * singleribosumscore( nseq1, nseq2, oseq1, oseq2, eff1, eff2, i, j ) * consweight_multi; } } } /* four-way consistency */ for( i=0; i<lgth1; i++ ) for( pairpt1=pairprob1[i]; pairpt1->bestpos!=-1; pairpt1++ ) { // if( pairprob1[i] == NULL ) continue; for( j=0; j<lgth2; j++ ) for( pairpt2=pairprob2[j]; pairpt2->bestpos!=-1; pairpt2++ ) { // fprintf( stderr, "i=%d, j=%d, pn1=%d, pn2=%d\n", i, j, pairpt1-pairprob1[i], pairpt2-pairprob2[j] ); // if( pairprob2[j] == NULL ) continue; uido = pairpt1->bestpos; ujdo = pairpt2->bestpos; prob = pairpt1->bestscore * pairpt2->bestscore; // prob = 1.0; // fprintf( stderr, "i=%d->uido=%d, j=%d->ujdo=%d\n", i, uido, j, ujdo ); // fprintf( stderr, "impmtx2[%d][%d] = %f\n", i, j, impmtx2[i][j] ); // if( i < uido && j > ujdo ) continue; // if( i > uido && j < ujdo ) continue; // posdistj = abs( ujdo-j ); // if( uido > -1 && ujdo > -1 ) if( uido > -1 && ujdo > -1 && ( ( i > uido && j > ujdo ) || ( i < uido && j < ujdo ) ) ) { { impmtx2[i][j] += MAX( 0, map[uido][ujdo] ) * consweight_rna * 600 * prob; // osoi } } } } for( i=0; i<lgth1; i++ ) for( j=0; j<lgth2; j++ ) { impmtx[i][j] += impmtx2[i][j]; // fprintf( stderr, "fastathreshold=%f, consweight_multi=%f, consweight_rna=%f\n", fastathreshold, consweight_multi, consweight_rna ); // impmtx[i][j] *= 0.5; } // impmtx[0][0] += 10000.0; // impmtx[lgth1-1][lgth2-1] += 10000.0; #if 0 fprintf( stdout, "#impmtx2 = \n" ); for( i=0; i<lgth1; i++ ) { for( j=0; j<lgth2; j++ ) { fprintf( stdout, "%d %d %f\n", i, j, impmtx2[i][j] ); } fprintf( stdout, "\n" ); } exit( 1 ); #endif } FreeCharMtx( useq1 ); FreeCharMtx( useq2 ); FreeCharMtx( oseq1 ); FreeCharMtx( oseq2 ); FreeCharMtx( oseq1r ); FreeCharMtx( oseq2r ); free( odir1 ); free( odir2 ); FreeFloatMtx( impmtx2 ); FreeFloatMtx( map ); FreeIntMtx( sgapmap1 ); FreeIntMtx( sgapmap2 ); FreeFloatMtx( tbppmtx ); for( i=0; i<lgth1; i++ ) free( pairprob1[i] ); for( i=0; i<lgth2; i++ ) free( pairprob2[i] ); free( pairprob1 ); free( pairprob2 ); }
int main( int argc, char *argv[] ) { static char com[10000]; static int *nlen; int left, right; int res; static char **name, **seq, **nogap; static int **gapmap; static int *order; int i, j; FILE *infp; RNApair ***pairprob; RNApair **alnpairprob; RNApair *pairprobpt; RNApair *pt; int *alnpairnum; float prob; int adpos; arguments( argc, argv ); #ifndef enablemultithread nthread = 0; #endif if( inputfile ) { infp = fopen( inputfile, "r" ); if( !infp ) { fprintf( stderr, "Cannot open %s\n", inputfile ); exit( 1 ); } } else infp = stdin; if( !whereismccaskillmea ) whereismccaskillmea = ""; getnumlen( infp ); rewind( infp ); if( dorp != 'd' ) { fprintf( stderr, "nuc only\n" ); exit( 1 ); } seq = AllocateCharMtx( njob, nlenmax*2+1 ); nogap = AllocateCharMtx( njob, nlenmax*2+1 ); gapmap = AllocateIntMtx( njob, nlenmax*2+1 ); order = AllocateIntVec( njob ); name = AllocateCharMtx( njob, B+1 ); nlen = AllocateIntVec( njob ); pairprob = (RNApair ***)calloc( njob, sizeof( RNApair ** ) ); alnpairprob = (RNApair **)calloc( nlenmax, sizeof( RNApair * ) ); alnpairnum = AllocateIntVec( nlenmax ); for( i=0; i<nlenmax; i++ ) alnpairnum[i] = 0; readData_pointer( infp, name, nlen, seq ); fclose( infp ); for( i=0; i<njob; i++ ) { pairprob[i] = (RNApair **)calloc( nlenmax, sizeof( RNApair * ) ); for( j=0; j<nlenmax; j++ ) { pairprob[i][j] = (RNApair *)calloc( 1, sizeof( RNApair ) ); pairprob[i][j][0].bestpos = -1; pairprob[i][j][0].bestscore = -1.0; } strcpy( nogap[i], seq[i] ); order[i] = i; } for( j=0; j<nlenmax; j++ ) { alnpairprob[j] = (RNApair *)calloc( 1, sizeof( RNApair ) ); alnpairprob[j][0].bestpos = -1; alnpairprob[j][0].bestscore = -1.0; } constants( njob, seq ); if( alg == 'G' ) fprintf( stderr, "Running DAFS (Sato et al. 2012; http://www.ncrna.org/).\n" ); else fprintf( stderr, "Running mxscarna with the mccaskill_mea mode.\n" ); #ifdef enablemultithread if( nthread > 0 ) { int jobpos; pthread_t *handle; pthread_mutex_t mutex; thread_arg_t *targ; jobpos = 0; targ = calloc( nthread, sizeof( thread_arg_t ) ); handle = calloc( nthread, sizeof( pthread_t ) ); pthread_mutex_init( &mutex, NULL ); for( i=0; i<nthread; i++ ) { targ[i].thread_no = i; targ[i].njob = njob; targ[i].jobpospt = &jobpos; targ[i].gapmap = gapmap; targ[i].nogap = nogap; targ[i].nlenmax = nlenmax; targ[i].pairprob = pairprob; targ[i].mutex = &mutex; // athread( targ ); pthread_create( handle+i, NULL, athread, (void *)(targ+i) ); } for( i=0; i<nthread; i++ ) { pthread_join( handle[i], NULL ); } pthread_mutex_destroy( &mutex ); for( i=0; i<njob; i++ ) { fprintf( stdout, ">%d\n", i ); outmccaskill( stdout, pairprob[i], nlenmax ); } } else #endif { for( i=0; i<njob; i++ ) { fprintf( stderr, "%d / %d\n", i+1, njob ); commongappick_record( 1, nogap+i, gapmap[i] ); infp = fopen( "_mccaskillinorg", "w" ); // fprintf( infp, ">in\n%s\n", nogap[i] ); fprintf( infp, ">in\n" ); write1seq( infp, nogap[i] ); fclose( infp ); system( "tr -d '\\r' < _mccaskillinorg > _mccaskillin" ); // for cygwin, wakaran if( alg == 'G' ) sprintf( com, "env PATH=%s dafs --mafft-out _mccaskillout _mccaskillin > _dum1 2>_dum", whereismccaskillmea ); else sprintf( com, "env PATH=%s mxscarnamod -m -writebpp _mccaskillin > _mccaskillout 2>_dum", whereismccaskillmea ); res = system( com ); if( res ) { fprintf( stderr, "ERROR IN mccaskill_mea\n" ); exit( 1 ); } infp = fopen( "_mccaskillout", "r" ); readrawmccaskill( infp, pairprob[i], nlenmax ); fclose( infp ); fprintf( stdout, ">%d\n", i ); outmccaskill( stdout, pairprob[i], nlenmax ); } } for( i=0; i<njob; i++ ) { for( j=0; j<nlen[i]; j++ ) for( pairprobpt=pairprob[i][j]; pairprobpt->bestpos!=-1; pairprobpt++ ) { left = gapmap[i][j]; right = gapmap[i][pairprobpt->bestpos]; prob = pairprobpt->bestscore; for( pt=alnpairprob[left]; pt->bestpos!=-1; pt++ ) if( pt->bestpos == right ) break; if( pt->bestpos == -1 ) { alnpairprob[left] = (RNApair *)realloc( alnpairprob[left], (alnpairnum[left]+2) * sizeof( RNApair ) ); adpos = alnpairnum[left]; alnpairnum[left]++; alnpairprob[left][adpos].bestscore = 0.0; alnpairprob[left][adpos].bestpos = right; alnpairprob[left][adpos+1].bestscore = -1.0; alnpairprob[left][adpos+1].bestpos = -1; pt = alnpairprob[left]+adpos; } else adpos = pt-alnpairprob[left]; pt->bestscore += prob; if( pt->bestpos != right ) { fprintf( stderr, "okashii!\n" ); exit( 1 ); } // fprintf( stderr, "adding %d-%d, %f\n", left, right, prob ); } } for( i=0; i<njob; i++ ) { for( j=0; j<nlenmax; j++ ) free( pairprob[i][j] ); free( pairprob[i] ); } free( pairprob ); for( j=0; j<nlenmax; j++ ) free( alnpairprob[j] ); free( alnpairprob ); free( alnpairnum ); fprintf( stderr, "%d thread(s)\n", nthread ); return( 0 ); #if 0 fprintf( stdout, "result=\n" ); for( i=0; i<nlenmax; i++ ) for( pairprobpt=alnpairprob[i]; pairprobpt->bestpos!=-1; pairprobpt++ ) { pairprobpt->bestscore /= (float)njob; left = i; right = pairprobpt->bestpos; prob = pairprobpt->bestscore; fprintf( stdout, "%d-%d, %f\n", left, right, prob ); } return( 0 ); #endif }
static void *athread( void *arg ) { thread_arg_t *targ = (thread_arg_t *)arg; int thread_no = targ->thread_no; int njob = targ->njob; int *jobpospt = targ->jobpospt; int **gapmap = targ->gapmap; char **nogap = targ->nogap; int nlenmax = targ->nlenmax; RNApair ***pairprob = targ->pairprob; int i, res; FILE *infp; char *com; char *dirname; dirname = calloc( 100, sizeof( char ) ); com = calloc( 1000, sizeof( char ) ); while( 1 ) { pthread_mutex_lock( targ->mutex ); i = *jobpospt; if( i == njob ) { pthread_mutex_unlock( targ->mutex ); return( NULL ); } *jobpospt = i+1; pthread_mutex_unlock( targ->mutex ); sprintf( dirname, "_%d", i ); sprintf( com, "rm -rf %s", dirname ); system( com ); sprintf( com, "mkdir %s", dirname ); system( com ); fprintf( stderr, "%d / %d (by thread %4d)\n", i+1, njob, thread_no ); commongappick_record( 1, nogap+i, gapmap[i] ); sprintf( com, "%s/_mccaskillinorg", dirname ); infp = fopen( com, "w" ); // fprintf( infp, ">in\n%s\n", nogap[i] ); fprintf( infp, ">in\n" ); write1seq( infp, nogap[i] ); fclose( infp ); sprintf( com, "tr -d '\\r' < %s/_mccaskillinorg > %s/_mccaskillin", dirname, dirname ); system( com ); // for cygwin, wakaran if( alg == 'G' ) sprintf( com, "cd %s; %s/dafs --mafft-out _mccaskillout _mccaskillin > _dum1 2>_dum", dirname, whereismccaskillmea ); else sprintf( com, "cd %s; %s/mxscarnamod -m -writebpp _mccaskillin > _mccaskillout 2>_dum", dirname, whereismccaskillmea ); res = system( com ); if( res ) { fprintf( stderr, "ERROR IN mccaskill_mea\n" ); exit( 1 ); } sprintf( com, "%s/_mccaskillout", dirname ); infp = fopen( com, "r" ); readrawmccaskill( infp, pairprob[i], nlenmax ); fclose( infp ); sprintf( com, "rm -rf %s > /dev/null 2>&1", dirname ); if( system( com ) ) { fprintf( stderr, "retrying to rmdir\n" ); // nanosleep( 100000 ); sleep( 1 ); system( com ); } } free( dirname ); free( com ); }
float Falign_localhom( char **seq1, char **seq2, double *eff1, double *eff2, int clus1, int clus2, int alloclen, LocalHom ***localhom, float *totalimpmatch, int *gapmap1, int *gapmap2, int *chudanpt, int chudanref, int *chudanres ) { // tditeration.c deha alloclen ha huhen nanode // prevalloclen ha iranai. int i, j, k, l, m, maxk; int nlen, nlen2, nlen4; static TLS int crossscoresize = 0; static TLS char **tmpseq1 = NULL; static TLS char **tmpseq2 = NULL; static TLS char **tmpptr1 = NULL; static TLS char **tmpptr2 = NULL; static TLS char **tmpres1 = NULL; static TLS char **tmpres2 = NULL; static TLS char **result1 = NULL; static TLS char **result2 = NULL; #if RND static TLS char **rndseq1 = NULL; static TLS char **rndseq2 = NULL; #endif static TLS Fukusosuu **seqVector1 = NULL; static TLS Fukusosuu **seqVector2 = NULL; static TLS Fukusosuu **naiseki = NULL; static TLS Fukusosuu *naisekiNoWa = NULL; static TLS double *soukan = NULL; static TLS double **crossscore = NULL; int nlentmp; static TLS int *kouho = NULL; static TLS Segment *segment = NULL; static TLS Segment *segment1 = NULL; static TLS Segment *segment2 = NULL; static TLS Segment **sortedseg1 = NULL; static TLS Segment **sortedseg2 = NULL; static TLS int *cut1 = NULL; static TLS int *cut2 = NULL; static TLS char *sgap1, *egap1, *sgap2, *egap2; static TLS int localalloclen = 0; int lag; int tmpint; int count, count0; int len1, len2; int totallen; float totalscore; float impmatch; extern Fukusosuu *AllocateFukusosuuVec(); extern Fukusosuu **AllocateFukusosuuMtx(); if( seq1 == NULL ) { if( result1 ) { // fprintf( stderr, "Freeing localarrays in Falign\n" ); localalloclen = 0; mymergesort( 0, 0, NULL ); alignableReagion( 0, 0, NULL, NULL, NULL, NULL, NULL ); fft( 0, NULL, 1 ); A__align( NULL, NULL, NULL, NULL, 0, 0, 0, NULL, NULL, NULL, NULL, NULL, NULL, NULL, 0, NULL, 0, 0 ); G__align11( NULL, NULL, 0, 0, 0 ); partA__align( NULL, NULL, NULL, NULL, 0, 0, 0, NULL, NULL, 0, 0, 0, 0, NULL, NULL, NULL, NULL, NULL, NULL, NULL, 0, NULL ); blockAlign2( NULL, NULL, NULL, NULL, NULL, NULL ); if( crossscore ) FreeDoubleMtx( crossscore ); FreeCharMtx( result1 ); FreeCharMtx( result2 ); FreeCharMtx( tmpres1 ); FreeCharMtx( tmpres2 ); FreeCharMtx( tmpseq1 ); FreeCharMtx( tmpseq2 ); free( sgap1 ); free( egap1 ); free( sgap2 ); free( egap2 ); free( kouho ); free( cut1 ); free( cut2 ); free( tmpptr1 ); free( tmpptr2 ); free( segment ); free( segment1 ); free( segment2 ); free( sortedseg1 ); free( sortedseg2 ); if( !kobetsubunkatsu ) { FreeFukusosuuMtx ( seqVector1 ); FreeFukusosuuMtx ( seqVector2 ); FreeFukusosuuVec( naisekiNoWa ); FreeFukusosuuMtx( naiseki ); FreeDoubleVec( soukan ); } } else { // fprintf( stderr, "Did not allocate localarrays in Falign\n" ); } return( 0.0 ); } len1 = strlen( seq1[0] ); len2 = strlen( seq2[0] ); nlentmp = MAX( len1, len2 ); nlen = 1; while( nlentmp >= nlen ) nlen <<= 1; #if 0 fprintf( stderr, "### nlen = %d\n", nlen ); #endif nlen2 = nlen/2; nlen4 = nlen2 / 2; #if DEBUG fprintf( stderr, "len1 = %d, len2 = %d\n", len1, len2 ); fprintf( stderr, "nlentmp = %d, nlen = %d\n", nlentmp, nlen ); #endif if( !localalloclen ) { sgap1 = AllocateCharVec( njob ); egap1 = AllocateCharVec( njob ); sgap2 = AllocateCharVec( njob ); egap2 = AllocateCharVec( njob ); kouho = AllocateIntVec( NKOUHO ); cut1 = AllocateIntVec( MAXSEG ); cut2 = AllocateIntVec( MAXSEG ); tmpptr1 = AllocateCharMtx( njob, 0 ); tmpptr2 = AllocateCharMtx( njob, 0 ); result1 = AllocateCharMtx( njob, alloclen ); result2 = AllocateCharMtx( njob, alloclen ); tmpres1 = AllocateCharMtx( njob, alloclen ); tmpres2 = AllocateCharMtx( njob, alloclen ); // crossscore = AllocateDoubleMtx( MAXSEG, MAXSEG ); segment = (Segment *)calloc( MAXSEG, sizeof( Segment ) ); segment1 = (Segment *)calloc( MAXSEG, sizeof( Segment ) ); segment2 = (Segment *)calloc( MAXSEG, sizeof( Segment ) ); sortedseg1 = (Segment **)calloc( MAXSEG, sizeof( Segment * ) ); sortedseg2 = (Segment **)calloc( MAXSEG, sizeof( Segment * ) ); if( !( segment && segment1 && segment2 && sortedseg1 && sortedseg2 ) ) ErrorExit( "Allocation error\n" ); if ( scoremtx == -1 ) n20or4or2 = 4; else if( fftscore == 1 ) n20or4or2 = 2; else n20or4or2 = 20; } if( localalloclen < nlen ) { if( localalloclen ) { #if 1 if( !kobetsubunkatsu ) { FreeFukusosuuMtx ( seqVector1 ); FreeFukusosuuMtx ( seqVector2 ); FreeFukusosuuVec( naisekiNoWa ); FreeFukusosuuMtx( naiseki ); FreeDoubleVec( soukan ); } FreeCharMtx( tmpseq1 ); FreeCharMtx( tmpseq2 ); #endif #if RND FreeCharMtx( rndseq1 ); FreeCharMtx( rndseq2 ); #endif } tmpseq1 = AllocateCharMtx( njob, nlen ); tmpseq2 = AllocateCharMtx( njob, nlen ); if( !kobetsubunkatsu ) { naisekiNoWa = AllocateFukusosuuVec( nlen ); naiseki = AllocateFukusosuuMtx( n20or4or2, nlen ); seqVector1 = AllocateFukusosuuMtx( n20or4or2+1, nlen+1 ); seqVector2 = AllocateFukusosuuMtx( n20or4or2+1, nlen+1 ); soukan = AllocateDoubleVec( nlen+1 ); } #if RND rndseq1 = AllocateCharMtx( njob, nlen ); rndseq2 = AllocateCharMtx( njob, nlen ); for( i=0; i<njob; i++ ) { generateRndSeq( rndseq1[i], nlen ); generateRndSeq( rndseq2[i], nlen ); } #endif localalloclen = nlen; } for( j=0; j<clus1; j++ ) strcpy( tmpseq1[j], seq1[j] ); for( j=0; j<clus2; j++ ) strcpy( tmpseq2[j], seq2[j] ); #if 0 fftfp = fopen( "input_of_Falign", "w" ); fprintf( fftfp, "nlen = %d\n", nlen ); fprintf( fftfp, "seq1: ( %d sequences ) \n", clus1 ); for( i=0; i<clus1; i++ ) fprintf( fftfp, "%s\n", seq1[i] ); fprintf( fftfp, "seq2: ( %d sequences ) \n", clus2 ); for( i=0; i<clus2; i++ ) fprintf( fftfp, "%s\n", seq2[i] ); fclose( fftfp ); system( "less input_of_Falign < /dev/tty > /dev/tty" ); #endif if( !kobetsubunkatsu ) { fprintf( stderr, "FFT ... " ); for( j=0; j<n20or4or2; j++ ) vec_init( seqVector1[j], nlen ); if( fftscore && scoremtx != -1 ) { for( i=0; i<clus1; i++ ) { seq_vec_2( seqVector1[0], polarity, eff1[i], tmpseq1[i] ); seq_vec_2( seqVector1[1], volume, eff1[i], tmpseq1[i] ); } } else { #if 0 for( i=0; i<clus1; i++ ) for( j=0; j<n20or4or2; j++ ) seq_vec( seqVector1[j], amino[j], eff1[i], tmpseq1[i] ); #else for( i=0; i<clus1; i++ ) seq_vec_3( seqVector1, eff1[i], tmpseq1[i] ); #endif } #if RND for( i=0; i<clus1; i++ ) { vec_init2( seqVector1, rndseq1[i], eff1[i], len1, nlen ); } #endif #if 0 fftfp = fopen( "seqVec", "w" ); fprintf( fftfp, "before transform\n" ); for( k=0; k<n20or4or2; k++ ) { fprintf( fftfp, "nlen=%d\n", nlen ); fprintf( fftfp, "%c\n", amino[k] ); for( l=0; l<nlen; l++ ) fprintf( fftfp, "%f %f\n", seqVector1[k][l].R, seqVector1[k][l].I ); } fclose( fftfp ); system( "less seqVec < /dev/tty > /dev/tty" ); #endif for( j=0; j<n20or4or2; j++ ) vec_init( seqVector2[j], nlen ); if( fftscore && scoremtx != -1 ) { for( i=0; i<clus2; i++ ) { seq_vec_2( seqVector2[0], polarity, eff2[i], tmpseq2[i] ); seq_vec_2( seqVector2[1], volume, eff2[i], tmpseq2[i] ); } } else { #if 0 for( i=0; i<clus2; i++ ) for( j=0; j<n20or4or2; j++ ) seq_vec( seqVector2[j], amino[j], eff2[i], tmpseq2[i] ); #else for( i=0; i<clus2; i++ ) seq_vec_3( seqVector2, eff2[i], tmpseq2[i] ); #endif } #if RND for( i=0; i<clus2; i++ ) { vec_init2( seqVector2, rndseq2[i], eff2[i], len2, nlen ); } #endif #if 0 fftfp = fopen( "seqVec2", "w" ); fprintf( fftfp, "before fft\n" ); for( k=0; k<n20or4or2; k++ ) { fprintf( fftfp, "%c\n", amino[k] ); for( l=0; l<nlen; l++ ) fprintf( fftfp, "%f %f\n", seqVector2[k][l].R, seqVector2[k][l].I ); } fclose( fftfp ); system( "less seqVec2 < /dev/tty > /dev/tty" ); #endif for( j=0; j<n20or4or2; j++ ) { fft( nlen, seqVector2[j], (j==0) ); fft( nlen, seqVector1[j], 0 ); } #if 0 fftfp = fopen( "seqVec2", "w" ); fprintf( fftfp, "#after fft\n" ); for( k=0; k<n20or4or2; k++ ) { fprintf( fftfp, "#%c\n", amino[k] ); for( l=0; l<nlen; l++ ) fprintf( fftfp, "%f %f\n", seqVector2[k][l].R, seqVector2[k][l].I ); } fclose( fftfp ); system( "less seqVec2 < /dev/tty > /dev/tty" ); #endif for( k=0; k<n20or4or2; k++ ) { for( l=0; l<nlen; l++ ) calcNaiseki( naiseki[k]+l, seqVector1[k]+l, seqVector2[k]+l ); } for( l=0; l<nlen; l++ ) { naisekiNoWa[l].R = 0.0; naisekiNoWa[l].I = 0.0; for( k=0; k<n20or4or2; k++ ) { naisekiNoWa[l].R += naiseki[k][l].R; naisekiNoWa[l].I += naiseki[k][l].I; } } #if 0 fftfp = fopen( "naisekiNoWa", "w" ); fprintf( fftfp, "#Before fft\n" ); for( l=0; l<nlen; l++ ) fprintf( fftfp, "%d %f %f\n", l, naisekiNoWa[l].R, naisekiNoWa[l].I ); fclose( fftfp ); system( "less naisekiNoWa < /dev/tty > /dev/tty " ); #endif fft( -nlen, naisekiNoWa, 0 ); for( m=0; m<=nlen2; m++ ) soukan[m] = naisekiNoWa[nlen2-m].R; for( m=nlen2+1; m<nlen; m++ ) soukan[m] = naisekiNoWa[nlen+nlen2-m].R; #if 0 fftfp = fopen( "naisekiNoWa", "w" ); fprintf( fftfp, "#After fft\n" ); for( l=0; l<nlen; l++ ) fprintf( fftfp, "%d %f\n", l, naisekiNoWa[l].R ); fclose( fftfp ); fftfp = fopen( "list.plot", "w" ); fprintf( fftfp, "plot 'naisekiNoWa'\npause -1" ); fclose( fftfp ); system( "/usr/bin/gnuplot list.plot &" ); #endif #if 0 fprintf( stderr, "frt write start\n" ); fftfp = fopen( "frt", "w" ); for( l=0; l<nlen; l++ ) fprintf( fftfp, "%d %f\n", l-nlen2, soukan[l] ); fclose( fftfp ); system( "less frt < /dev/tty > /dev/tty" ); #if 0 fftfp = fopen( "list.plot", "w" ); fprintf( fftfp, "plot 'frt'\n pause +1" ); fclose( fftfp ); system( "/usr/bin/gnuplot list.plot" ); #endif #endif getKouho( kouho, NKOUHO, soukan, nlen ); #if 0 for( i=0; i<NKOUHO; i++ ) { fprintf( stderr, "kouho[%d] = %d\n", i, kouho[i] ); } #endif } #if KEIKA fprintf( stderr, "Searching anchors ... " ); #endif count = 0; #define CAND 0 #if CAND fftfp = fopen( "cand", "w" ); fclose( fftfp ); #endif if( kobetsubunkatsu ) { maxk = 1; kouho[0] = 0; } else { maxk = NKOUHO; } for( k=0; k<maxk; k++ ) { lag = kouho[k]; zurasu2( lag, clus1, clus2, seq1, seq2, tmpptr1, tmpptr2 ); #if CAND fftfp = fopen( "cand", "a" ); fprintf( fftfp, "Candidate No.%d lag = %d\n", k+1, lag ); fprintf( fftfp, "%s\n", tmpptr1[0] ); fprintf( fftfp, "%s\n", tmpptr2[0] ); fclose( fftfp ); #endif tmpint = alignableReagion( clus1, clus2, tmpptr1, tmpptr2, eff1, eff2, segment+count ); if( count+tmpint > MAXSEG -3 ) ErrorExit( "TOO MANY SEGMENTS.\n" ); while( tmpint-- > 0 ) { if( lag > 0 ) { segment1[count].start = segment[count].start ; segment1[count].end = segment[count].end ; segment1[count].center = segment[count].center; segment1[count].score = segment[count].score; segment2[count].start = segment[count].start + lag; segment2[count].end = segment[count].end + lag; segment2[count].center = segment[count].center + lag; segment2[count].score = segment[count].score ; } else { segment1[count].start = segment[count].start - lag; segment1[count].end = segment[count].end - lag; segment1[count].center = segment[count].center - lag; segment1[count].score = segment[count].score ; segment2[count].start = segment[count].start ; segment2[count].end = segment[count].end ; segment2[count].center = segment[count].center; segment2[count].score = segment[count].score ; } #if 0 fftfp = fopen( "cand", "a" ); fprintf( fftfp, "Goukaku=%dko\n", tmpint ); fprintf( fftfp, "in 1 %d\n", segment1[count].center ); fprintf( fftfp, "in 2 %d\n", segment2[count].center ); fclose( fftfp ); #endif segment1[count].pair = &segment2[count]; segment2[count].pair = &segment1[count]; count++; #if 0 fprintf( stderr, "count=%d\n", count ); #endif } } #if 1 if( !kobetsubunkatsu ) fprintf( stderr, "%d segments found\n", count ); #endif if( !count && fftNoAnchStop ) ErrorExit( "Cannot detect anchor!" ); #if 0 fftfp = fopen( "fft", "a" ); fprintf( fftfp, "RESULT before sort:\n" ); for( l=0; l<count; l++ ) { fprintf( fftfp, "cut[%d]=%d, ", l, segment1[l].center ); fprintf( fftfp, "%d score = %f\n", segment2[l].center, segment1[l].score ); } fclose( fftfp ); #endif #if KEIKA fprintf( stderr, "Aligning anchors ... " ); #endif for( i=0; i<count; i++ ) { sortedseg1[i] = &segment1[i]; sortedseg2[i] = &segment2[i]; } #if 0 tmpsort( count, sortedseg1 ); tmpsort( count, sortedseg2 ); qsort( sortedseg1, count, sizeof( Segment * ), segcmp ); qsort( sortedseg2, count, sizeof( Segment * ), segcmp ); #else mymergesort( 0, count-1, sortedseg1 ); mymergesort( 0, count-1, sortedseg2 ); #endif for( i=0; i<count; i++ ) sortedseg1[i]->number = i; for( i=0; i<count; i++ ) sortedseg2[i]->number = i; if( kobetsubunkatsu ) { for( i=0; i<count; i++ ) { cut1[i+1] = sortedseg1[i]->center; cut2[i+1] = sortedseg2[i]->center; } cut1[0] = 0; cut2[0] = 0; cut1[count+1] = len1; cut2[count+1] = len2; count += 2; } else { if( crossscoresize < count+2 ) { crossscoresize = count+2; #if 1 fprintf( stderr, "######allocating crossscore, size = %d\n", crossscoresize ); #endif if( crossscore ) FreeDoubleMtx( crossscore ); crossscore = AllocateDoubleMtx( crossscoresize, crossscoresize ); } for( i=0; i<count+2; i++ ) for( j=0; j<count+2; j++ ) crossscore[i][j] = 0.0; for( i=0; i<count; i++ ) { crossscore[segment1[i].number+1][segment1[i].pair->number+1] = segment1[i].score; cut1[i+1] = sortedseg1[i]->center; cut2[i+1] = sortedseg2[i]->center; } #if DEBUG fprintf( stderr, "AFTER SORT\n" ); for( i=0; i<count; i++ ) fprintf( stderr, "%d, %d\n", segment1[i].start, segment2[i].start ); #endif crossscore[0][0] = 10000000.0; cut1[0] = 0; cut2[0] = 0; crossscore[count+1][count+1] = 10000000.0; cut1[count+1] = len1; cut2[count+1] = len2; count += 2; count0 = count; blockAlign2( cut1, cut2, sortedseg1, sortedseg2, crossscore, &count ); if( count0 > count ) { #if 0 fprintf( stderr, "\7 REPEAT!? \n" ); #else fprintf( stderr, "REPEAT!? \n" ); #endif if( fftRepeatStop ) exit( 1 ); } #if KEIKA else fprintf( stderr, "done\n" ); #endif } #if 0 fftfp = fopen( "fft", "a" ); fprintf( fftfp, "RESULT after sort:\n" ); for( l=0; l<count; l++ ) { fprintf( fftfp, "cut[%d]=%d, ", l, segment1[l].center ); fprintf( fftfp, "%d\n", segment2[l].center ); } fclose( fftfp ); #endif #if 0 fftfp = fopen( "fft", "a" ); fprintf( fftfp, "RESULT after sort:\n" ); for( l=0; l<count; l++ ) { fprintf( fftfp, "cut : %d %d\n", cut1[l], cut2[l] ); } fclose( fftfp ); #endif #if KEIKA fprintf( trap_g, "Devided to %d segments\n", count-1 ); fprintf( trap_g, "%d %d forg\n", MIN( clus1, clus2 ), count-1 ); #endif totallen = 0; for( j=0; j<clus1; j++ ) result1[j][0] = 0; for( j=0; j<clus2; j++ ) result2[j][0] = 0; totalscore = 0.0; *totalimpmatch = 0.0; for( i=0; i<count-1; i++ ) { #if DEBUG fprintf( stderr, "DP %03d / %03d %4d to ", i+1, count-1, totallen ); #else #if KEIKA fprintf( stderr, "DP %03d / %03d\r", i+1, count-1 ); #endif #endif if( cut1[i] ) { getkyokaigap( sgap1, seq1, cut1[i]-1, clus1 ); getkyokaigap( sgap2, seq2, cut2[i]-1, clus2 ); } else { for( j=0; j<clus1; j++ ) sgap1[j] = 'o'; for( j=0; j<clus2; j++ ) sgap2[j] = 'o'; } if( cut1[i+1] != len1 ) { getkyokaigap( egap1, seq1, cut1[i+1], clus1 ); getkyokaigap( egap2, seq2, cut2[i+1], clus2 ); } else { for( j=0; j<clus1; j++ ) egap1[j] = 'o'; for( j=0; j<clus2; j++ ) egap2[j] = 'o'; } for( j=0; j<clus1; j++ ) { strncpy( tmpres1[j], seq1[j]+cut1[i], cut1[i+1]-cut1[i] ); tmpres1[j][cut1[i+1]-cut1[i]] = 0; } if( kobetsubunkatsu ) commongappick_record( clus1, tmpres1, gapmap1 ); for( j=0; j<clus2; j++ ) { strncpy( tmpres2[j], seq2[j]+cut2[i], cut2[i+1]-cut2[i] ); tmpres2[j][cut2[i+1]-cut2[i]] = 0; } if( kobetsubunkatsu ) commongappick_record( clus2, tmpres2, gapmap2 ); #if 0 fprintf( stderr, "count = %d\n", count ); fprintf( stderr, "### reg1 = %d-%d\n", cut1[i], cut1[i+1]-1 ); fprintf( stderr, "### reg2 = %d-%d\n", cut2[i], cut2[i+1]-1 ); #endif switch( alg ) { case( 'a' ): totalscore += Aalign( tmpres1, tmpres2, eff1, eff2, clus1, clus2, alloclen ); break; case( 'Q' ): totalscore += partQ__align( tmpres1, tmpres2, eff1, eff2, clus1, clus2, alloclen, localhom, &impmatch, cut1[i], cut1[i+1]-1, cut2[i], cut2[i+1]-1, gapmap1, gapmap2, sgap1, sgap2, egap1, egap2 ); *totalimpmatch += impmatch; // fprintf( stderr, "*totalimpmatch in Falign_localhom = %f\n", *totalimpmatch ); break; case( 'A' ): totalscore += partA__align( tmpres1, tmpres2, eff1, eff2, clus1, clus2, alloclen, localhom, &impmatch, cut1[i], cut1[i+1]-1, cut2[i], cut2[i+1]-1, gapmap1, gapmap2, sgap1, sgap2, egap1, egap2, chudanpt, chudanref, chudanres ); *totalimpmatch += impmatch; // fprintf( stderr, "*totalimpmatch in Falign_localhom = %f\n", *totalimpmatch ); break; default: fprintf( stderr, "alg = %c\n", alg ); ErrorExit( "ERROR IN SOURCE FILE Falign.c" ); break; } #ifdef enablemultithread if( chudanres && *chudanres ) { // fprintf( stderr, "\n\n## CHUUDAN!!! at Falign_localhom\n" ); return( -1.0 ); } #endif nlen = strlen( tmpres1[0] ); if( totallen + nlen > alloclen ) { fprintf( stderr, "totallen=%d + nlen=%d > alloclen = %d\n", totallen, nlen, alloclen ); ErrorExit( "LENGTH OVER in Falign\n " ); } for( j=0; j<clus1; j++ ) strcat( result1[j], tmpres1[j] ); for( j=0; j<clus2; j++ ) strcat( result2[j], tmpres2[j] ); totallen += nlen; #if 0 fprintf( stderr, "%4d\r", totallen ); fprintf( stderr, "\n\n" ); for( j=0; j<clus1; j++ ) { fprintf( stderr, "%s\n", tmpres1[j] ); } fprintf( stderr, "-------\n" ); for( j=0; j<clus2; j++ ) { fprintf( stderr, "%s\n", tmpres2[j] ); } #endif } #if KEIKA fprintf( stderr, "DP ... done \n" ); #endif for( j=0; j<clus1; j++ ) strcpy( seq1[j], result1[j] ); for( j=0; j<clus2; j++ ) strcpy( seq2[j], result2[j] ); #if 0 for( j=0; j<clus1; j++ ) { fprintf( stderr, "%s\n", result1[j] ); } fprintf( stderr, "- - - - - - - - - - -\n" ); for( j=0; j<clus2; j++ ) { fprintf( stderr, "%s\n", result2[j] ); } #endif return( totalscore ); }