regex - 从 fasta 文件生成随机子集序列

Question

向全世界的 Perl 大师们问好。

我在编程方面遇到了另一个麻烦。我正在编写一个程序，该程序从具有特定输入编号的蛋白质组 fasta 文件中选择随机序列。

一般的 fasta 文件如下所示:

>seq_ID_1 descriptions etc ASDGDSAHSAHASDFRHGSDHSDGEWTSHSDHDSHFSDGSGASGADGHHAH ASDSADGDASHDASHSAREWAWGDASHASGASGASGSDGASDGDSAHSHAS SFASGDASGDSSDFDSFSDFSD

>seq_ID_2 descriptions etc ASDGDSAHSAHASDFRHGSDHSDGEWTSHSDHDSHFSDGSGASGADGHHAH ASDSADGDASHDASHSAREWAWGDASHASGASGASG

and so on.......

字母代表氨基酸肽。

所以我有一个包含 1000 个序列的 fasta 文件，我想检索其中的 63.21%，即 632.1 个序列。但是序列不能是 float ，所以如果它超过 0.5 我想四舍五入，如果小于 0.5 我想四舍五入。

这是我生成随机序列子集的代码，但它的工作效率稍差。

#!/usr/bin/perl

#Selecting 63.21% of random sequnces from a proteom file.
use strict;
use warnings;
use List::Util qw(shuffle);

#Give the first argument as a proteom file.
if (@ARGV != 1)
{
    print "Invalid arguments\n";
    print "Usage: perl randseq.pl [proteom_file]";
    exit(0);
}

my $FILE = $ARGV[0];
my $i = 0;
my %protseq = {};
my $nIdx = 0;

#Extraction and counting of the all headers from a proteom file.
open(LIST,$FILE);
open(TEMP1, ">temp1");
while (my $line = <LIST>){
    chomp $line;
    if ($line =~ />(\S+) (.+)/){
        $i++;
        print TEMP1 $1,"\n";
    }
}
close(LIST);
close(TEMP1);

#Selection of random headers for generating a random subset of the proteom file.
my $GET_LINES = RoundToInt ($i*0.6321);

my @line_starts;
open(my $FH,'<','temp1');
open(TEMP2, ">temp2");

do {
     push @line_starts, tell $FH
} while ( <$FH> );

my $count = @line_starts;

my @shuffled_starts = (shuffle @line_starts)[1..$GET_LINES+1];

for my $start ( @shuffled_starts ) {

     seek $FH, $start, 0
         or die "Unable to seek to line - $!\n";

     print TEMP2 scalar <$FH>;
}
close(TEMP2);

#Assigning the sequence data to randomly generated header file.
open(DATA,'<','temp2');
while(my $line = <DATA>)
{
    chomp($line);
    $line =~ s/[\t\s]//g;
    if($line =~ /^([^\s]+)/)
    {
        $protseq{$1}++;
    }
}
close(DATA);

open(DATA, "$FILE");
open(OUT, ">random_seqs.fasta");
while(my $line = <DATA>)
{
    chomp($line);
    if($line =~ /^>([^\s]+)/)
    {
        if($protseq{$1} ne "")
        {

            $nIdx = 1;
            print OUT "$line\n";
        }
        else
        {
            $nIdx = 0;
        }
    }
    else
    {
        if($nIdx == 1)
        {
            print OUT "$line\n";
        }
    }
}
close(DATA);
close(OUT);

#subroutine for rounding
sub RoundToInt {
  int($_[0] + .5 * ($_[0] <=> 0));
}

system("erase temp1");
system("erase temp2");
exit;

但是，它有时会给出正确数量的序列，有时会多出一个序列。我怎样才能摆脱那个……请问有什么想法吗？

或者更短的代码？

在这里您可以获得一个75酵母蛋白质组文件。 [http://www.peroxisomedb.org/Download/Saccharomyces_cerevisiae.fas][1]

希望我能尽快解决这个问题...:(

Answer 1

您的方法看起来不错，只是不必要地复杂了。我会这样做:

use strict;
use warnings;

# usage: randseq.pl [fraction] < input.fasta > output.fasta
my $fraction = (@ARGV ? shift : 0.6321);

# Collect input lines into an array of sequences:
my @sequences;
while (<>) {
    # A leading > starts a new sequence. (The "\" is only there to
    # avoid confusing the Stack Overflow syntax highlighting.)
    push @sequences, [] if /^\>/;
    push @{ $sequences[-1] }, $_;
}

# Calculate how many sequences we want:
my $n = @sequences;
my $k = int( $n * $fraction + 0.5 );
warn "Selecting $k out of $n sequences (", 100 * $k / $n, "%).\n";

# Do a partial Fisher-Yates shuffle to select $k random sequences out of $n:
foreach my $i (0 .. $k-1) {
    my $j = $i + int rand($n-$i);
    @sequences[$i,$j] = @sequences[$j,$i];
}

# Print the output:
print @$_ for @sequences[0 .. $k-1];

请注意，此代码将输入文件的全部内容读入内存。如果输入文件太大，而你只想要它的一小部分，可以使用 reservoir sampling从任意大的集合中选择 k 个随机序列而无需保留更多:

use strict;
use warnings;

my $k = (@ARGV ? shift : 632);  # sample size: need to know this in advance

# Use reservoir sampling to select $k random sequences:
my @samples;
my $n = 0;  # total number of sequences read
my $i;      # index of current sequence
while (<>) {
    if (/^\>/) {
        # Select a random sequence from 0 to $n-1 to replace:
        $i = int rand ++$n;
        # Save all samples until we've accumulated $k of them:
        $samples[$n-1] = $samples[$i] if $n <= $k;
        # Only actually store the new sequence if it's one of the $k first ones:
        $samples[$i] = [] if $i < $k;
    }
    push @{ $samples[$i] }, $_ if $i < $k;
}

warn "Only read $n < $k sequences, selected all.\n" if $n < $k;
warn "Selected $k out of $n sequences (", 100 * $k / $n, "%).\n" if $n >= $k;

# Print sampled sequences:
print @$_ for @samples;

但是，如果您真的想要输入序列的某个部分，您需要首先在文件的单独传递中对它们进行计数。

作为副作用，上述两个程序也统一打乱采样序列。 (事实上​​ ，我故意调整了水库采样算法，使 n 和 k 的所有值的洗牌统一。)如果你不想那样，你总是可以在打印序列之前，根据您喜欢的任何标准对序列进行排序。