template<class T> inline static Matrix<T>
smooth (const Matrix<T>& y, const size_t& span = 3, const INTERP::Method& method = INTERP::AKIMA) {
typedef typename TypeTraits<T>::RT RT;
Loess<T> l ((RT)span/(RT)size(y,0));
Matrix<T> ret = l.lowess(y);
return ret;
}
/** main loop */
void run()
{
set<string> all_chromosomes;
/** data shuttle, will be used by the pileup function */
PileupShuttle shuttle;
shuttle.owner=this;
/* loop over the exon */
size_t exon_index=0;
while( exon_index< this->exons.size())
{
WHERE(exon_index << "/"<<this->exons.size() );
shuttle.exon = this->exons.at(exon_index);
/* delete exon size<=25 */
if(shuttle.exon->size()<= this->minimum_exon_size)
{
this->exons.erase(exon_index);
continue;
}
//get gc% for this segement
shuttle.exon->gc_percent=gcPercentAt(
shuttle.exon->segment.chrom.c_str(),
shuttle.exon->segment.start,
shuttle.exon->segment.end
);
shuttle.exon->coverage.resize(this->bamFiles.size(),0);
/** loop over each sample */
for(shuttle.bam_index=0;
shuttle.bam_index< this->bamFiles.size();
++shuttle.bam_index
)
{
shuttle.bamFile = this->bamFiles.at(shuttle.bam_index);
int32_t tid=shuttle.bamFile->findTidByName(shuttle.exon->segment.chrom.c_str());
if(tid<0) THROW("chromosome name not indexed in tid.");
bam_plbuf_t *buf= ::bam_plbuf_init(scan_depth_func, &shuttle);
::bam_fetch(
shuttle.bamFile->bamPtr(),
shuttle.bamFile->bamIndex(),
tid,
shuttle.exon->segment.start,
shuttle.exon->segment.end,
buf,
BamFile2::fetch_func
);
bam_plbuf_push(0, buf);
bam_plbuf_destroy(buf);
//normalize to exon length
shuttle.exon->coverage[shuttle.bam_index]/=(double)( shuttle.exon->size() );
}
/* min-depth */
if(shuttle.exon->min_depth()<=0)
{
this->exons.erase(exon_index);
continue;
}
/* mediane_depthh */
if(shuttle.exon->median_depth()<10)
{
this->exons.erase(exon_index);
continue;
}
all_chromosomes.insert(shuttle.exon->segment.chrom);
++exon_index;
}/* end of loop over exons */
vector<Exon*> sort_on_gc;
for(size_t i=0;i< this->exons.size();++i)
{
sort_on_gc.push_back(this->exons.at(i));
}
/* sort on GC% */
std::sort(sort_on_gc.begin(),sort_on_gc.end(),cmp_on_gc);
fstream os("jeter.tar",ios::out);
Tar tarball(os);
std::string prefix("redondepth");
std::string filename;
FILE* gnuplotout= safe_tmpfile();
/* reset gnuplot */
fprintf(gnuplotout,
"set term postscript\n"
"set output \"redondepth.ps\"\n"
);
/* loop over each individual */
for(size_t sampleidx1=0;sampleidx1< this->bamFiles.size();++sampleidx1)
{
char folder_name[50];
sprintf(folder_name,"SAMPLE%d",(int)(1+sampleidx1));
string prefix2(prefix);
prefix2.append("/").append(folder_name);
/* initialize */
for(size_t i=0;i< sort_on_gc.size();++i)
{
Exon* exon=sort_on_gc.at(i);
exon->normalized_coverage.clear();
}
/* loop over the other individuals */
for(size_t sampleidx2=0;sampleidx2< this->bamFiles.size();++sampleidx2)
{
/* skip if same individual */
if(sampleidx1==sampleidx2) continue;
vector<double> x;
vector<double> y;
/** loop over each exon and build lowess X/Y */
for(size_t i=0;i< sort_on_gc.size();++i)
{
Exon* exon=sort_on_gc.at(i);
x.push_back(exon->gc_percent);
y.push_back(log2(exon->coverage[sampleidx1]/exon->coverage[sampleidx2]));
}
auto_ptr<vector<double> > y_prime=loessAlgo.lowess(&(x.front()),&(y.front()),x.size());
/* loop over all exons and push normalized ratio*/
for(size_t i=0;i< sort_on_gc.size();++i)
{
Exon* exon=sort_on_gc.at(i);
double l=log2(exon->coverage[sampleidx1]/exon->coverage[sampleidx2]);
exon->normalized_coverage.push_back(l-y_prime->at(i));
}
}
for(std::set<string>::iterator r=all_chromosomes.begin();
r!=all_chromosomes.end();
++r)
{
FILE* outdat = safe_tmpfile();
for(size_t i=0;i< this->exons.size();++i)
{
Exon* exon=this->exons.at(i);
if(exon->segment.chrom.compare(*r)!=0) continue;
fprintf(
outdat,
"%d\t%f\n",
exon->segment.start,
exon->median_normalized_coverage()
);
}
string filename_dat(prefix2);
filename_dat.append("/").append(*r).append(".dat");
tarball.putFile(outdat,filename_dat.c_str());
fclose(outdat);
fprintf(gnuplotout,"set title \"SAMPLE-%d %s\"\n",(int)(1+sampleidx1),r->c_str());
fprintf(gnuplotout,"set xlabel \"Position\"\n");
fprintf(gnuplotout,"set ylabel \"Median depth\"\n");
fprintf(gnuplotout,"set yrange [-2:2]\n");
fprintf(gnuplotout,"plot \"%s/%s.dat\" using 1:2 notitle\n",folder_name,r->c_str());
}
}
filename.assign(prefix).append("/gnuplot.txt");
tarball.putFile(gnuplotout,filename.c_str());
tarball.finish();
fclose(gnuplotout);
os.flush();
os.close();
}
