根据已知的参考基因组和GTF创建适合自己分析的基因组和GTF

#!/usr/bin/perl
open(FR,"10.fa")||die;
open(FW,">10_cut.fa")||die;
 
while(<FR>) {
chomp;
if ($_=~/(>\d+)/)
{
print FW "$1\n";
}
else {print FW "$_\n";}
}
close FR;
close FW;

#!/usr/bin/perl
open(FR,"10.fa")||die;
open(FW,">10_add_chr.fa")||die;
 
while(<FR>) {
chomp;
if($_ =~/^>/)
{
$_ =~ s/>/>chr/g;
print FW "$_\n"
}
else {print FW "$_\n";}
}
close FR;
close FW;

#!/usr/bin/perl
open(FW,"Macaca_mulatta.MMUL_1.73.gtf")||die;
open(FM,">DEL_Macaca_mulatta.MMUL_1.73_del_un.gtf")||die;
while(<FW>)
{
chomp;
if($_=~/^1099*/)
{
print FM ""
}
else
{
print FM "$_\n"
}
}
close FW;
close FM;

#!/usr/bin/perl
#first chr numble must add chr.
open(FR,"DEL_Macaca_mulatta.MMUL_1.73_del_un.gtf")||die;
open(FW,">Add_chr_Macaca_mulatta.MMUL_1.73_del_un.gtf")||die;
 
while(<FR>) {
chomp;
if($_=~/^\d/ or $_=~/^X/ or $_=~/^MT/)
{print FW "chr$_\n"}
else {print FW "\n"}
}
close FR;
close FW;

for i in `seq 20`
do
grep '^'${i}''$'\t' Macaca_mulatta.MMUL_1.73.gtf >${i}.gtf
done
 
grep '^X'$'\t' Macaca_mulatta.MMUL_1.73.gtf >X.gtf
grep '^MT'$'\t' Macaca_mulatta.MMUL_1.73.gtf >MT.gtf
 
cat *.gtf >Macaca_mulatta_1.73_final.gtf #合并每个GTF；