一个 fasta 序列格式化小脚本

一个 fasta 小工具，主要用来格式化fasta序列的，也可以实现 fasta 核酸序列转为 pep 蛋白序列。

#!/usr/bin/perl -w
use strict;
use Getopt::Long;

my ($infile, $outfile);
my ($alignLen, $cds2pep, $remove, $caseType);
my ($version, $help);

GetOptions(
    "infile|i=s"  => \$infile,
    # "outfile|o:s" => \$outfile,
    "caseType|c:s"=> \$caseType,
    "alignLen:i"  => \$alignLen,
    "cds2pep"   => \$cds2pep,
    "version"   => \$version,
    "help|?|h"  => \$help
);

&usage if (!defined $infile);
$alignLen ||= 60;
$caseType ||= "";

open FA, "<$infile" || die "Erro! Can't open file $infile\n";
$/ = ">";;
while(){
    chomp;
    my ($info, $seq) = split (/\n/, $_, 2);
    # my $id = &keepPart($info, 1 ,' ');  # split id line with blank
    my $out;

    if (defined $alignLen){
       $seq =~ s/\n//g;
       $out = &out_fasta($seq, $alignLen);
       # $out .= "\n";
    }
    
    if ($caseType =~ /Uper/i){
        $out = uc($out);
        # print ">$id\n$out";
    }elsif( $caseType =~ /low/i){
        $out = lc($out);
        # print ">$id\n$out";
    }else{
        $out = $out;
        # print ">$id\n$out";
    }

    if (defined $cds2pep){
        $out = &cds2pep($out);
        # print ">$id\n$out"
    }

    print ">$info\n$out\n";
}
close FA;
$/="\n";

sub keepPart{
    my ($info, $col, $spp) = @_;
    my @tmp = split /$spp/, $info;
    my $out = $tmp[$col-1];
    return $out;
}

sub out_fasta{
    my ($seq, $num) = @_;
    my $len = length $seq;
    $seq =~ s/([A-Za-z]{$num})/$1\n/g;
    chop($seq) unless $len % $num;
    return $seq;
}

sub cds2pep{
    my $cds_fasta = shift @_;
    my $p = &code();
    my $pep_fasta;

    for(my $i = 0; $i < length $cds_fasta; $i += 3){
        my $codon = uc(substr($cds_fasta,$i,3));
        last if (length $codon < 3);
        $pep_fasta .= exists $p->{$codon} ? $p->{$codon} : "X";
    }
    return $pep_fasta;
}

sub code {
    my $codon = {
        'GCA' => 'A', 'GCC' => 'A', 'GCG' => 'A', 'GCT' => 'A',                               # Alanine
        'TGC' => 'C', 'TGT' => 'C',                                                           # Cysteine
        'GAC' => 'D', 'GAT' => 'D',                                                           # Aspartic Aci
        'GAA' => 'E', 'GAG' => 'E',                                                           # Glutamic Aci
        'TTC' => 'F', 'TTT' => 'F',                                                           # Phenylalanin
        'GGA' => 'G', 'GGC' => 'G', 'GGG' => 'G', 'GGT' => 'G',                               # Glycine
        'CAC' => 'H', 'CAT' => 'H',                                                           # Histidine
        'ATA' => 'I', 'ATC' => 'I', 'ATT' => 'I',                                             # Isoleucine
        'AAA' => 'K', 'AAG' => 'K',                                                           # Lysine
        'CTA' => 'L', 'CTC' => 'L', 'CTG' => 'L', 'CTT' => 'L', 'TTA' => 'L', 'TTG' => 'L',   # Leucine
        'ATG' => 'M',                                                                         # Methionine
        'AAC' => 'N', 'AAT' => 'N',                                                           # Asparagine
        'CCA' => 'P', 'CCC' => 'P', 'CCG' => 'P', 'CCT' => 'P',                               # Proline
        'CAA' => 'Q', 'CAG' => 'Q',                                                           # Glutamine
        'CGA' => 'R', 'CGC' => 'R', 'CGG' => 'R', 'CGT' => 'R', 'AGA' => 'R', 'AGG' => 'R',   # Arginine
        'TCA' => 'S', 'TCC' => 'S', 'TCG' => 'S', 'TCT' => 'S', 'AGC' => 'S', 'AGT' => 'S',   # Serine
        'ACA' => 'T', 'ACC' => 'T', 'ACG' => 'T', 'ACT' => 'T',                               # Threonine
        'GTA' => 'V', 'GTC' => 'V', 'GTG' => 'V', 'GTT' => 'V',                               # Valine
        'TGG' => 'W',                                                                         # Tryptophan
        'TAC' => 'Y', 'TAT' => 'Y',                                                           # Tyrosine
        'TAA' => 'U', 'TAG' => 'U', 'TGA' => 'U'                                              # Stop
    };
    return $codon;
}

sub usage {
    print STDERR<

还可以根据需要加点小功能。