BEGIN{
    f="/data/results/reference/mmu/mm9/mRNA-stranded/Mus_musculus.NCBIM37.64-toMM9.headers";
    while (getline<f){
	en=$1;split($2,a,"=");split($3,b,"=");
	if (length(a[2])==0){print "ERR no gene "$0;}
	if (length(b[2])==0){print "ERR no CDS "$0;}
	split(b[2],c,"-");
	gi[en]=a[2];cdss[en]=c[1];cdse[en]=c[2];
	ok[en]=1;
    }
    close(f);
    FS="\t";
}{
    en=$1;
    if (ok[en]==1){
	s=$2;e=$3;ncl++;iu5=0;iu3=0;icds=0;hit[en]++;
	if (e<cdss[en]) {u5++;iu5=1;nu5[en]++;}
	if (s>cdse[en]) {u3++;iu3=1;nu3[en]++;}
	if ( (s>=cdss[en])&&(e<=cdse[en]) ) {cds++;icds=1;ncds[en]++;}
#	print cdss[en]" "cdse[en]" "iu5" "icds" "iu3" "$0;
    } else{
	print "ERR unk ENS "$0;
    }
}END{
print  "nclusters "ncl"  in5UTR "u5"  "100.0*u5/ncl"  inCDS "u5"  "100.0*cds/ncl"  in3UTR "u3"  "100.0*u3/ncl
for (en in hit){
    nhit++;
    thit+=hit[en];
    tu3+=nu3[en];
    tu5+=nu5[en];
    tcds+=ncds[en];        
    thit2+=hit[en]*hit[en];
    tu32+=nu3[en]*nu3[en];
    tu52+=nu5[en]*nu5[en];
    tcds2+=ncds[en]*ncds[en];
}
print "transcripts_with_clusters "nhit"  avg_n_clusters_per_transcript "1.0*thit/nhit"  avg_n_clusters_per_5UTR "1.0*tu5/nhit"  avg_n_clusters_per_CDS "1.0*tcds/nhit"  avg_n_clusters_per_3UTR "1.0*tu3/nhit
 }