patscan.pl

这是我用来搜索序列中目标段的代码～ 注：其中有些地址

use strict;
use warnings;

#open the source data files
my $inPath = "D:\\patscan\\input\\";
my $outPath = "D:\\patscan\\output\\";
open(SOURCE_FILE, $inPath."3UTR.Mus_nr.dat") || die "cannot open source file!\n";
open(OUT_FILE,">".$outPath."result.dat") || die "cannot creat output file!\n";
my $count1 = 0;
my $count2 = 0;

#get records and check if there is identical sequence
while(my $record = get_next_record(SOURCE_FILE)) {
	#print $record,"\n";
	my $microSeq = get_micro_seq($record);
        #print $microSeq;
        my $ID = get_ID($record);
        while($microSeq =~ /CACTGCC/ig){
        	my $pos = length $`;
                $count1++;
        	print OUT_FILE "CACTGCC ","ID = ",$ID," ","Position = ",$pos," ","\n";
                print "CACTGCC ","ID = ",$ID," ","Position = ",$pos," ","\n";
        };
        while($microSeq =~ /ACTGCCT/ig){
        	my $pos = length $`;
                $count2++;
        	print OUT_FILE  "ACTGCCT ","ID = ",$ID," ","Position = ",$pos," ","\n";
                print  "ACTGCCT ","ID = ",$ID," ","Position = ",$pos," ","\n";
        };
}
print OUT_FILE "Num of CACTGCC = $count1\n","Num of ACTGCCT = $count2", "\n";
print "Num of CACTGCC = $count1\n","Num of ACTGCCT = $count2", "\n";
print "success!\n";
print OUT_FILE "success!\n";
close(SOURCE_FILE);
close(OUT_FILE);
exit;

sub get_next_record {
	my $fileHandle = shift;    #@_中究竟是什么？
	my $record = '';
	my $saveInputSep = $/;
	$/ = "//\n";
	$record = <$fileHandle>;
	$/ = $saveInputSep;
	return $record;
}

sub get_micro_seq{
	my $record = shift;       #为什么这些地方都使用shift
        $record =~ /^SQ.*\n((.*\n)+)/m;
        $preSeq = $1;
        $preSeq =~ s/\s//g;     #去除数据中的空格
        $preSeq =~ s/\d//g;     #去除数据中的数字
        return $preSeq;         #此时的数据中只有序列，没有其他的干扰

}

sub get_ID{
	my $record = shift;
        $record =~/^ID   (\S*)\s*/m;
       	my $ID = $1;
        return $ID;
}

#By Jun Chen and Yi Dongliang