/** \brief named c-tor normalizing given value to [0;1] range.
* \param v value to normalize.
* \return valid object representing similarity.
*/
static const Similarity normalize(const double v)
{
if(v<0)
return Similarity(0);
if(1<v)
return Similarity(1);
return Similarity(v);
}
//1 means it is a gap, 0 otherwise
//assume the length of Cors, Type must be of length this->inseqs.size()
float Alignment::pos_score(MtxE* mtx,int* Cors,bool* Type){
float score_sum=0;
for(unsigned int i=0; i < this->inseqs.size();i++){
if(Type[i] == 1){
score_sum += this->Gap*((int)this->inseqs.size()-i-1);
continue;
}
for(unsigned int j=i+1;j < this->inseqs.size();j++){
if(Type[j] == 1)
score_sum += this->Gap;
else{
float Percentage=Similarity(this->inseqs[i][Cors[i]-1],this->inseqs[j][Cors[j]-1]);
if(FloatEqual(Percentage,1.0) == true ){
score_sum+=this->Match;
}else if(FloatEqual(Percentage,0.9) == true){
score_sum+=this->Identical;
}else if(Percentage >= 0.66 ){
score_sum+=this->Similar;
}else{
score_sum+=this->Mismatch;
}
}
}
}
return (2.0*score_sum)/(((int)this->inseqs.size())*((int)this->inseqs.size()-1));
}
void DaoxDebugger::Matching( QList<DaoVmCode> & x, QList<DaoVmCode> & y )
{
align1.clear();
align2.clear();
Similarity( x, y );
Matching( x.size(), y.size() );
int i, j, k, m, n = align1.size();
#if 0
//printf( "x: %i; y: %i\n", x.size(), y.size() );
for(i=0; i<x.size(); i++) DaoVmCode_Print( x[i], NULL );
for(i=0; i<y.size(); i++) DaoVmCode_Print( y[i], NULL );
#endif
for(i=0; i<n; i++){
DaoCnode xnode, ynode;
int a1 = align1[i];
int a2 = align2[i];
#if 0
//printf( "%3i: %3i %3i\n", i, a1, a2 );
if( a1 >=0 ) DaoVmCode_Print( x[a1], NULL );
if( a2 >=0 ) DaoVmCode_Print( y[a2], NULL );
#endif
if( a1 < 0 || a2 < 0 ) continue;
memset( & xnode, 0, sizeof(DaoCnode) );
memset( & ynode, 0, sizeof(DaoCnode) );
DaoCnode_InitOperands( & xnode, & x[a1] );
DaoCnode_InitOperands( & ynode, & y[a2] );
switch( xnode.type ){
case DAO_OP_NONE :
break;
case DAO_OP_SINGLE :
if( regmap.find( xnode.first ) == regmap.end() ) regmap[ xnode.first ] = ynode.first;
break;
case DAO_OP_PAIR :
if( regmap.find( xnode.first ) == regmap.end() ) regmap[ xnode.first ] = ynode.first;
if( regmap.find( xnode.second ) == regmap.end() ) regmap[ xnode.second ] = ynode.second;
break;
case DAO_OP_TRIPLE :
if( regmap.find( xnode.first ) == regmap.end() ) regmap[ xnode.first ] = ynode.first;
if( regmap.find( xnode.second ) == regmap.end() ) regmap[ xnode.second ] = ynode.second;
if( regmap.find( xnode.third ) == regmap.end() ) regmap[ xnode.third ] = ynode.third;
break;
case DAO_OP_RANGE :
case DAO_OP_RANGE2 :
for(i=xnode.first; i<=xnode.second; i++){
if( regmap.find( i ) == regmap.end() ) regmap[i] = ynode.first + i - xnode.first;
}
if( xnode.type == DAO_OP_RANGE2 ){
if( regmap.find( xnode.third ) == regmap.end() ) regmap[ xnode.third ] = ynode.third;
}
break;
}
}
}
float MAlignment::align(){
profile.push_back(seqs[0]);
float FinalScore;
int total_length=seqs[0].size();
for(unsigned int i=1;i < seqs.size();i++){
total_length+=seqs[i].size();
/*******Calculate the TABLE********/
int ydim=profile[0].size()+1;
int xdim=seqs[i].size()+1;
float score[xdim*ydim];
int trace[xdim*ydim]; // to record traces for finding the aligned sequences 1 for gap in x,2 for match,3 for gap in y
for(int l=0; l < ydim;l++){
score[l*xdim]=(float)l*gap;
trace[l*xdim]=3;
}
for(int l=0; l < xdim;l++){
score[l]=(float)l*gap;
trace[l]=1;
}
for(int j=1; j <xdim;j++){
for(int k=1; k <ydim;k++){
float temp=0.0;
score[k*xdim+j]=score[(k-1)*xdim+j]+gap;//initialize min with adding gap in y direction
trace[k*xdim+j]=3;
for(unsigned int m=0;m < profile.size();m++){
float Percentage=Similarity(profile[m][k-1],seqs[i][j-1]);
if(FloatEqual(Percentage,1.0) == true){
temp=temp+score[(k-1)*xdim+j-1]+identical;
//cout<<"Identical "<<profile[m][k-1]<<","<<seqs[i][j-1]<<" Score:"<<temp<<endl;
}else if(FloatEqual(Percentage,0.9) == true){
temp=temp+score[(k-1)*xdim+j-1]+verysimilar;
//cout<<"Almost "<<profile[m][k-1]<<","<<seqs[i][j-1]<<" Score:"<<temp<<endl;
}else if(Percentage >= 0.66 ){
//cout<<"Temp now is "<<temp<<endl;
temp=temp+score[(k-1)*xdim+j-1]+similar;
//cout<<"Similar "<<profile[m][k-1]<<","<<seqs[i][j-1]<<" Score:"<<temp<<endl;
}else if(Percentage < 0.0 ){
temp=temp+score[(k-1)*xdim+j-1]+gap;
//cout<<"MatchToGap "<<profile[m][k-1]<<","<<seqs[i][j-1]<<" Score:"<<temp<<endl;
}else{
temp=temp+score[(k-1)*xdim+j-1]+mismatch;
//cout<<"Mismatch "<<profile[m][k-1]<<","<<seqs[i][j-1]<<" Score:"<<temp<<endl;
}
}
temp=temp/float(profile.size());
//cout<<"Finally the score is "<<temp<<endl;
if(temp < score[k*xdim+j]){
//cout<<temp << " is smaller than "<< score[k*xdim+j] <<endl;
score[k*xdim+j]=temp;
trace[k*xdim+j]=2;
}
temp=score[k*xdim+j-1]+gap; //adding gap in x direction
if(temp < score[k*xdim+j]){
//cout<<temp << " is smaller than "<< score[k*xdim+j] <<endl;
score[k*xdim+j]=temp;
trace[k*xdim+j]=1;
}
}
}
//cout<<"Finish Scoring?"<<endl;
/************** Output score matrix*************
for(int k=ydim-1; k >=0;k--){
if( k >= 1){
cout<<profile[0][k-1];
}else{
cout<<"-";
}
for(int j=0; j <xdim;j++){
cout<<"\t"<<score[k*xdim+j];
}
cout<<endl;
}
cout<<"\t"<<"-";
for(unsigned int j=0; j < seqs[i].size();j++){
cout<<"\t"<<seqs[i][j];
}
cout<<endl;
cout<<"Final Score :"<<score[xdim*ydim-1]<<endl;
*************************************************/
/************** Out put matrix
for(int l=ydim-1; l >=0;l--){
for(int m=0; m <xdim;m++){
cout<<"\t"<<trace[l*xdim+m];
}
cout<<endl;
}
*/
/**Trace back find the alignment*/
int xInd=xdim-1;
int yInd=ydim-1;
while(xInd > 0 || yInd > 0 ){
//cout <<"Checking "<<xInd<<","<<yInd<<endl;
if(xInd-1 >=0 && trace[yInd*xdim+xInd] == 1){
for(unsigned int m=0;m < profile.size();m++){
if(alignment.size() > m){
alignment[m].insert(alignment[m].begin(),"-");
}else{
vector<string> temp;
temp.push_back("-");
alignment.push_back(temp);
}
}
if(alignment.size() > profile.size() ){
alignment[int(profile.size())].insert(alignment[int(profile.size())].begin(),seqs[i][xInd-1]);
}else{
vector<string> temp;
temp.push_back(seqs[i][xInd-1]);
alignment.push_back(temp);
}
xInd=xInd-1;
continue;
}
if(yInd-1 >=0 && trace[yInd*xdim+xInd] == 3){
for(unsigned int m=0;m < profile.size();m++){
if(alignment.size() > m){
alignment[m].insert(alignment[m].begin(),profile[m][yInd-1]);
}else{
vector<string> temp;
temp.push_back(profile[m][yInd-1]);
alignment.push_back(temp);
}
}
if(alignment.size() > profile.size()){
alignment[int(profile.size())].insert(alignment[int(profile.size())].begin(),"-");
}else{
vector<string> temp;
temp.push_back("-");
alignment.push_back(temp);
}
yInd=yInd-1;
continue;
}
if(trace[yInd*xdim+xInd] == 2){
for(unsigned int m=0;m < profile.size();m++){
if(alignment.size() > m){
alignment[m].insert(alignment[m].begin(),profile[m][yInd-1]);
}else{
vector<string> temp;
temp.push_back(profile[m][yInd-1]);
alignment.push_back(temp);
}
}
if(alignment.size() > profile.size()){
alignment[int(profile.size())].insert(alignment[int(profile.size())].begin(),seqs[i][xInd-1]);
}else{
vector<string> temp;
temp.push_back(seqs[i][xInd-1]);
alignment.push_back(temp);
}
xInd=xInd-1;
yInd=yInd-1;
}
}
//~ while (i > 0)
//~ {
//~ AlignmentA <- Ai + AlignmentA
//~ AlignmentB <- "-" + AlignmentB
//~ i <- i - 1
//~ }
//~ while (j > 0)
//~ {
//~ AlignmentA <- "-" + AlignmentA
//~ AlignmentB <- Bj + AlignmentB
//~ j <- j - 1
//~ }
UpdateProfile();
FinalScore=score[xdim*ydim-1];
}
return (FinalScore)/(int)(profile[0].size())+2;
}
/**
* \brief Main function.
*/
int main(int argc, char **argv)
{
char *s,*t;
int size,sizes,sizet;
int i,j,k,P;
int cond;
int *simi,res,Paux;
int *a,*b;
FILE *f,*f2;
fpos_t filepos;
int my_rank,set;
struct timeval ini, fi;
struct timezone tz;
bsp_begin(atoi(argv[1]));
size = atoi(argv[1]);
f=fopen(argv[2],"r");
if (f==NULL) Exit("Error: File %s not found\n",argv[2]);
fscanf(f,"%d",&sizes);
if (sizes%size != 0)
Exit("Error: The sequences have to have multiple of "
"processes quantity size");
f2=fopen(argv[3],"r");
if (f2==NULL) Exit("Error: File %s not found\n",argv[3]);
fscanf(f2,"%d",&sizet);
if (bsp_pid() == 0)
if (sizet%size != 0)
Exit("Error: The sequences have to have multiple of "
"processes quantity size");
P = atoi(argv[4]);
if (bsp_pid() == 0)
printf("align %d %s %s %d\n",size,argv[2],argv[3],P);
sizes /= size;
sizet /= size;
s = (char*) malloc (sizes*sizeof(char));
t = (char*) malloc (sizet*sizeof(char));
if (s == NULL || t == NULL)
Exit("No memory\n");
a = (int*)malloc ((sizet+1)*sizeof(int));
b = (int*)malloc ((sizes+1)*sizeof(int));
if (a == NULL || b == NULL)
Exit("No memory\n");
if (bsp_pid() == size-1)
{
simi = (int*) malloc(P*sizeof(int));
if (simi == NULL) Exit("No memory\n");
}
Paux = 0;
bsp_push_reg(s,sizes*sizeof(char));
bsp_push_reg(b,(sizes+1)*sizeof(int));
bsp_push_reg(&filepos,sizeof(long int));
bsp_push_reg(&i,sizeof(int));
bsp_sync();
gettimeofday(&ini,&tz);
for (k = 0; k < P*size + size -1; k++)
{
if (k >= bsp_pid() && k <= P*size+bsp_pid()-1)
cond = 1;
else
cond = 0;
set = 0;
if (cond==1 && (k-bsp_pid())%size == 0)/*start of a reading*/
{
if (bsp_pid() == 0 && k < size);
else if (bsp_pid() == 0)
{
bsp_get(size-1,&filepos,0,&filepos,sizeof(long int));
}
else
{
bsp_get(bsp_pid()-1,&filepos,0,&filepos,sizeof(long int));
}
set = 1;
}
bsp_sync();
if (cond==1 && (k-bsp_pid())%size == 0)/*start of a reading*/
{
if (set == 1) fsetpos(f2,&filepos);
for (i = 0; i < sizet; i++)
{
fscanf(f2,"%c",&t[i]);
if (t[i] == 'A' ||t[i] == 'T' ||t[i] == 'C' ||t[i] == 'G');
else
{
if (t[i] == EOF) Exit("Error: End of file reached without"
"read all sequence in %s\n",argv[3]);
i--;
}
}
fgetpos(f2,&filepos);
for (i = 0; i <= sizet; i++)
a[i] = (i+bsp_pid()*sizet)*gap;
}
if (cond==1)
{
if (bsp_pid() == 0)
{
for (i = 0; i < sizes; i++)
{
fscanf(f,"%c",&s[i]);
if (s[i] == 'A' ||s[i] == 'T' ||s[i] == 'C' ||s[i] == 'G');
else
{
if (s[i] == EOF) Exit("Error: End of file reached without"
"read all sequence in %s\n",argv[2]);
i--;
}
}
for (j = 0; j <= sizes; j++)
b[j] = (j + (k%size)*sizes)*gap;
}
res = Similarity (s, sizes, t, sizet, a, b);
if (bsp_pid() == size-1 && (k-bsp_pid()+1)%size == 0)
{
simi[Paux++] = res;
}
}
if (cond)
{
if (bsp_pid() != size -1)
{
bsp_put(bsp_pid()+1,s,s,0,sizes*sizeof(char));
bsp_put(bsp_pid()+1,b,b,0,(sizes+1)*sizeof(int));
}
}
bsp_sync();
}
gettimeofday(&fi,&tz);
printf("process %d ended\n",bsp_pid());
fclose(f);
fclose(f2);
if (bsp_pid() == size-1)
{
printf("Similarities: ");
for (i = 0; i < P; i++)
printf("%d ",simi[i]);
printf("\n");
}
if (bsp_pid() == 0)
{
printf("Computation time: %f\n", (fi.tv_sec - ini.tv_sec + (double)(fi.tv_usec -
ini.tv_usec)/1000000)/60);
}
bsp_pop_reg(&filepos);
bsp_pop_reg(b);
bsp_pop_reg(s);
bsp_sync();
return 0;
}