The entire text file looks something similar like this

3 9606 34 ACADM 187960098 NP_001120800.1

5 9606 37 ACADVL 4557235 NP_000009.1

6 9615 489421 ACAT1 73955189 XP_546539.2

with the line number as the hash table index

huh, why not just use an array?

my @a;
while (<>) {
   chomp;
   my @rec = split;
   push @a, [ @rec[???, ???, ???] ];
}
[download]

i should store all the desired columns

You keep saying you only want certain columns, yet you don't say which. Again, just use the index of the columns you want for the question marks.

while($line=<>) {
  ($key,$num1,$num2,$string,$num3,$stringnum) = split(/\s/,$line);
  $somehash{"$key"}{"$num1"}{"$num2"}{"$string"}{"$num3"}= $stringnum;
}
[download]

that puts the data into a "hash", but probably not what you want.
whether you use a hash or array structure largely depends on the data available and the logic/processing required. sequential processing and lack of a random access key lends itself to an array structure. when you have a good logical random access key (not a record sequence number) and need to access the records non-sequentially, use a hash. a hash structure, or even a mix of hash and array structure may be suitable. but exactly what structure do you want? both approaches could be out the window if you have millions of records in the file, whereby some much smarter arrangement would be required to achieve the logic/processing required.
speaking of which..what is the required logic/processing for these records?
the hardest line to type correctly is: stty erase ^H