int eseqclusterData::getOTU(float dist,earray<eintarray>& otus,int size)
{
eintarray otuarr;
int i,j;
otuarr.init(size);
for (i=0; i<size; ++i)
otuarr[i]=i;
// point at the seq we merged with
for (i=0; i<mergearr.size() && mergearr[i].dist>=dist; ++i)
otuarr[mergearr[i].y]=mergearr[i].x;
eintarray otuind;
otuind.init(otuarr.size(),-1);
// translate all the merged seq ids to the last seq id of the otu
for (i=0; i<otuarr.size(); ++i){
for (j=otuarr[i]; j!=otuarr[j]; j=otuarr[j]);
otuarr[i]=j;
if (otuind[j]==-1){
otuind[j]=otus.size();
otus.add(eintarray());
}
}
for (i=0; i<otuarr.size(); ++i)
otus[otuind[otuarr[i]]].add(i);
return(otus.size());
}
void eseqclustersingle::init(INDTYPE count,const estr& ofilename,const estr& seqsfile,const earray<ebasicarray<INDTYPE> >& dupslist) {
ofile.open(ofilename,"w");
ofile.write("# seqsfile: "+seqsfile+"\n");
ofile.write("# OTU_count Merge_distance Merged_OTU_id1 Merged_OTU_id2\n");
long i,j;
mergecount=0;
scount.reserve(count);
scluster.reserve(count);
smerge.reserve(count);
incluster.reserve(count);
for (i=0; i<count; ++i){
scount.add(1);
scluster.add(i);
smerge.add(-1);
incluster.add(list<INDTYPE>());
incluster[i].push_back(i);
}
for (i=0; i<dupslist.size(); ++i){
for (j=1; j<dupslist[i].size(); ++j){
++mergecount;
ofile.write(estr(scluster.size()-mergecount)+" 1.0 "+dupslist[i][0]+" "+dupslist[i][j]+"\n");
clusterData.mergearr.add(eseqdist(dupslist[i][0],dupslist[i][j],1.0));
}
}
cout << "# initializing cluster with: "<< count<< " seqs" << endl;
}
void eseqclusteravg::init(INDTYPE count,const estr& filename,const estr& seqsfile,const earray<ebasicarray<INDTYPE> >& dupslist,float _thres,float (_fdist)(const estr&,const estr&,int),estrarray& _seqarr,int _seqlen)
{
thres=_thres;
seqarr=&_seqarr;
seqlen=_seqlen;
fdist=_fdist;
ofile.open(filename,"w");
ofile.write("# seqsfile: "+seqsfile+"\n");
ofile.write("# OTU_count Merge_distance Merged_OTU_id1 Merged_OTU_id2\n");
long i,j;
incmaxdist=1.0;
incmaxit=smatrix.end();
cf=0;
lastdist=0.0;
scount.reserve(count);
scluster.reserve(count);
smerge.reserve(count);
inter.reserve(count);
incluster.reserve(count);
mergecount=0;
for (i=0; i<count; ++i){
scount.add(1);
scluster.add(i);
smerge.add(-1);
incluster.add(list<INDTYPE>());
incluster[i].push_back(i);
inter.add(list<INDTYPE>());
}
for (i=0; i<dupslist.size(); ++i){
for (j=1; j<dupslist[i].size(); ++j){
++mergecount;
ofile.write(estr(scluster.size()-mergecount)+" 1.0 "+dupslist[i][0]+" "+dupslist[i][j]+"\n");
clusterData.mergearr.add(eseqdist(dupslist[i][0],dupslist[i][j],1.0));
}
}
cout << "# initializing cluster with: "<< count<< " seqs" << endl;
cout << "# initializing smatrix with: " << (long)(count)*(long)(count)/20000l/2l<< " elements" << endl;
smatrix.reserve((long)(count)*(long)(count)/20000l/2l);
// cout << "# smatrix._hashitems = " << smatrix._hashitems << endl;
}