void GroupFromAnnotation::GetGeneMap(String path)
{
IFILE genemap;
genemap = ifopen(mapFile,"r");
if(genemap==NULL)
{
if(mapFile=="../data/refFlat_hg19.txt")
{
mapFile += ".gz";
genemap = ifopen(mapFile,"r");
if(genemap==NULL)
{
int loc = path.Find("bin");
if(loc!=-1)
{
mapFile = path.Left(loc-1);
mapFile += "/data/refFlat_hg19.txt";
}
else
{
mapFile += "../data/refFlat_hg19.txt";
}
genemap = ifopen(mapFile,"r");
}
if(genemap==NULL)
{
mapFile += ".gz";
genemap = ifopen(mapFile,"r");
}
if(genemap==NULL)
error("Cannot open gene mapping file %s.\n",mapFile.c_str());
}
else
error("Cannot open gene mapping file %s.\n",mapFile.c_str());
}
StringIntHash GeneLocHash;
StringArray strand;
int gene_idx =0;
while(!ifeof(genemap))
{
String buffer;
buffer.ReadLine(genemap);
StringArray record;
record.AddTokens(buffer,"\t");
int loc = GeneLocHash.Integer(record[0]);
if(loc==-1)
{
GeneLocHash.SetInteger(record[0],gene_idx);
//save chr, start and end positions
StringArray gene_chr;
if(record[2][2]=='r' || record[2][2]=='R')
record[2] = record[2].SubStr(3);
gene_chr.AddTokens(record[2],"_,;.");
if(gene_chr[0].Find("Un")!=-1)
continue;
/*
if(ChrLocHash.Integer(gene_chr[0])==-1)
{
chr_count++;
unique_chr.Push(gene_chr[0]);
ChrLocHash.SetInteger(gene_chr[0],chr_count);
}
*/
chr.Push(gene_chr[0]);
//printf("%d\t%s\t%s\n",idx,record[0].c_str(),gene_chr[0].c_str());
start_pos.Push(record[4].AsInteger());
end_pos.Push(record[5].AsInteger());
strand.Push(record[3]);
genename.Push(record[0]);
gene_idx++;
}
else
{
//get the current chr
StringArray gene_chr;
if(record[2][2]=='r' || record[2][2]=='R')
record[2] = record[2].SubStr(3);
gene_chr.AddTokens(record[2],"_,;.");
if(gene_chr[0].Find("Un")!=-1)
continue;
//check if strand and chr are consistent with previous record
if(chr[loc]!=gene_chr[0])
//if(strand[loc]!=record[3] || chr[loc]!=gene_chr[0])
// printf("Gene %s in %s has multiple records in different chromosome or strand.\n",record[0].c_str(),mapFile.c_str());
continue;
//update start and end position
if(record[4].AsInteger()<start_pos[loc])
start_pos[loc] = record[4].AsInteger();
if(record[5].AsInteger()>end_pos[loc])
end_pos[loc] = record[5].AsInteger();
}
}
ifclose(genemap);
//ifclose(genemap);
chr_idx.Index(chr);
String chr_=chr[chr_idx[0]];
for(int i=1;i<chr.Length();i++)
{
if(chr[chr_idx[i]]!=chr_)
{
ChrStartHash.SetInteger(chr[chr_idx[i]],i);
ChrEndHash.SetInteger(chr_,i-1);
chr_ = chr[chr_idx[i]];
}
}
}
