Remove particular columns of a matrix

soblanc has asked for the wisdom of the Perl Monks concerning the following question:

Hello,

I can't manage to remove particular columns of my multidimentional dataframe (matrix). It's not strictly duplicates.

Let's consider the matrix below :

my @matrix = (
    [t1 t1 t2 t2 t2], <- transcripts
    [a1 a2 a1 a1 a2], <- alleles
    [intron intron UTR_CG UTR UTR], <- locations
);
[download]

In fact, I would like to delete colums where I have "UTR" alone for a given transcript and allele, but ONLY when I already have a column with "UTR_CG" for the same given transcript and allele. For instance, the resulting table would be :

my @matrix = (
    [t1 t1 t2 t2],
    [a1 a2 a1 a2],
    [intron intron UTR_CG UTR],
);
[download]

Thank you so much in advance!

Comment on Remove particular columns of a matrix Select or Download Code

Replies are listed 'Best First'.
Re: Remove particular columns of a matrix by tybalt89 (Monsignor) on Sep 08, 2022 at 20:34 UTC
Transpose is your friend... #!/usr/bin/perl use strict; # https://perlmonks.org/?node_id=11146766 use warnings; use List::AllUtils qw( zip_by ); my @matrix = ( [qw(t1 t1 t2 t2 t2)], # <- transcripts [qw(a1 a2 a1 a1 a2)], # <- alleles [qw(intron intron UTR_CG UTR UTR)], # <- locations ); use Data::Dump 'dd'; dd 'original matrix', \@matrix; my @t = zip_by { [ @_ ] } @matrix; use Data::Dump 'dd'; dd '@t', \@t; my %cg = map { $_->[0] . "\n" . $_->[1] => 1 } grep { $_->[2] eq 'UTR_ +CG' } @t; use Data::Dump 'dd'; dd '%cg', \%cg; @t = grep { $_->[2] ne 'UTR' or not $cg{ $_->[0] . "\n" . $_->[1] } } +@t; use Data::Dump 'dd'; dd 'modified @t', \@t; my @finalmatrix = zip_by { [ @_ ] } @t; use Data::Dump 'dd'; dd 'final matrix', \@finalmatrix; [download] Outputs: ( "original matrix", [ ["t1", "t1", "t2", "t2", "t2"], ["a1", "a2", "a1", "a1", "a2"], ["intron", "intron", "UTR_CG", "UTR", "UTR"], ], ) ( "\@t", [ ["t1", "a1", "intron"], ["t1", "a2", "intron"], ["t2", "a1", "UTR_CG"], ["t2", "a1", "UTR"], ["t2", "a2", "UTR"], ], ) ("%cg", { "t2\na1" => 1 }) ( "modified \@t", [ ["t1", "a1", "intron"], ["t1", "a2", "intron"], ["t2", "a1", "UTR_CG"], ["t2", "a2", "UTR"], ], ) ( "final matrix", [ ["t1", "t1", "t2", "t2"], ["a1", "a2", "a1", "a2"], ["intron", "intron", "UTR_CG", "UTR"], ], ) [download]	[reply] [d/l] [select]
Re: Remove particular columns of a matrix by kcott (Archbishop) on Sep 08, 2022 at 17:29 UTC
G'day soblanc, "I can't manage to remove particular columns of my multidimentional dataframe (matrix)." You haven't shown your code nor indicated what's specifically causing difficulties. My best hint would be to use splice(); although, I've no idea if you're already using that. See "How do I post a question effectively?" and "Short, Self-Contained, Correct Example". — Ken	[reply]
Re: Remove particular columns of a matrix by Marshall (Canon) on Sep 09, 2022 at 03:07 UTC
Some index utils in List:MoreUtils are useful; use strict; use warnings; use List::MoreUtils qw(firstidx indexes); use Data::Dump qw(pp); my @matrix = ( [qw(t1 t1 t2 t2 t2)], [qw(a1 a2 a1 a1 a2)], [qw(intron intron UTR_CG UTR UTR)], ); die "No UTR_CG column!" unless ( (firstidx {$_ eq 'UTR_CG'}@{$matrix[- +1]}) > -1); my @utr_cols = indexes { $_ eq 'UTR' } @{$matrix[-1]}; die "Must have 2 or more UTR columns!" unless (@utr_cols >= 2); shift @utr_cols; #save first seen UTR column delete @{$matrix[$_]}[@utr_cols] for (0..@matrix-1); pp \@matrix; __END__ [ ["t1", "t1", "t2", "t2"], ["a1", "a2", "a1", "a1"], ["intron", "intron", "UTR_CG", "UTR"], ] [download]	[reply] [d/l]
Re: Remove particular columns of a matrix by LanX (Saint) on Sep 08, 2022 at 18:41 UTC
my advice use a `grep` to find all indices matching your fuzzy UTR criteria (?) and assign it to an array @indices apply a delete on a array-slice, something like `delete @{$matrix[$_]}[@indices] for 0..2` Like kcott I'd rather prefer you to show some more effort. The pseudo code you demonstrated doesn't even compile. Cheers Rolf _{(addicted to the Perl Programming Language :) Wikisyntax for the Monastery}	[reply] [d/l]