Try this to see if it's what you need:

use strict;
use warnings;
use Data::Dumper;

my @array = ();
while (my $line = <$ifh>) {
  push(@array, [ (split(/\t/, $line))[0,3,7] ]);
}

print Dumper(\@array);
[download]

Update: BTW, in your spec, you say the 1st, 4th, and 8th column, but in your output data you show the 1st, 3rd and 8th columns. Which is it?

Update: I see that you have updated your root node to reflect the corrected indices. Generally, it is considered to be polite to indicate (via an Update: such as this) that you have changed the wording/context of your node. Otherwise, it makes the respondents look like *they* made the mistake.

Is this question different than Parsing information from text file? Please don't post basically the same question multiple times.

open my $fh, "<", "file_name_here.txt" or die $!;
my @data;
my $header = <$fh>;
while(<$fh>) {
  chomp;
  my @row = split "\t";
  push @data, [ @row[0, 3, 7] ];
}
close($fh) or die $!;
[download]

If your TSV is more like a CSV - if it has quoting for example - the you should use Text::CSV_XS instead of split. But if it's just raw data and raw tabs this should work fine.

-sam

my @array;
while (<>){
   next if $.<2; #skip 1st line
   push @array, [(split / /, $_)[0,2,7]];
}
[download]