如何按 Perl 中的一列分类

Question

I'm studying about Gene programing by perl. It is little difficult for me using Perl. I am sorry about i'm not good at English.

I want to categorize by one column in perl.

This is my file. And file name is Annuum.v.2.1.gff3

PGAv.1.6.scaffold1  PROTEIN gene    909002  910083  .   +   .   ID=CA.PGAv.1.6.scaffold1.1
PGAv.1.6.scaffold1  PROTEIN mRNA    909002  910083  .+ ID=TC.CA.PGAv.1.6.scaffold1.1;Parent=CA.PGAv.1.6.scaffold1.1

PGAv.1.6.scaffold1  PROTEIN exon    909002  909168  .   +   0   Parent=TC.CA.PGAv.1.6.scaffold1.1
PGAv.1.6.scaffold1  PROTEIN CDS 909002  909168  .   + 0 Parent=TC.CA.PGAv.1.6.scaffold1.1

PGAv.1.6.scaffold1  PROTEIN exon    909759  910083  .   +   1   Parent=TC.CA.PGAv.1.6.scaffold1.1
PGAv.1.6.scaffold1  PROTEIN CDS 909759  910083  .   +   1   Parent=TC.CA.PGAv.1.6.scaffold1.1

PGAv.1.6.scaffold1  ISGAP   gene    930723  931169  783 +   .   ID=CA.PGAv.1.6.scaffold1.2
PGAv.1.6.scaffold1  ISGAP   mRNA    930723  931169  783 +   .   ID=TC.CA.PGAv.1.6.scaffold1.2;Parent=CA.PGAv.1.6.scaffold1.2

PGAv.1.6.scaffold1  ISGAP   exon    930723  931169  .   +   .   Parent=TC.CA.PGAv.1.6.scaffold1.2

PGAv.1.6.scaffold1  ISGAP   CDS 930723  931169  .   +   .   Parent=TC.CA.PGAv.1.6.scaffold1.2

I want to categorize by second column and i want to show the count number and ID. Like this below. I forgot sth. These are all genes.

PROTEIN number      CA.PGAv.1.6.scaffold1.1, CA.PGAv.1.6.scaffold1.3, ...

ISGAP   number          CA.PGAv.1.6.scaffold1.2, CA.PGAv.1.6.scaffold1.26, ...

Please help me. Thanks.

Answer 1

We can use an implicit loop and autosplit mode together with a hash of the categories and arrays for each category. This gives:

#!/usr/bin/perl -anF\t
next unless /ID=([^;]*)\n/; # process only lines with only ID
$categories{$F[1]} = 1;     # remember category (field 1)
push @{$F[1]}, $1;          # add item to the category's array
END { for (keys %categories)
      { $number = @{$_};    # number of items in category
        print "$_\t$number\t", join(", ", @{$_}), $/
      }
    }