Perl Script To Make Statistics Of Mirna Abundances For Many Samples

Question

[biolab]

HI everyone,

I am making statistics of mirna abundances for many samples. Below is an example for input file.

SAMPLE    MIR    ABUNDANCE
sample1   mir1   30
sample1   mir3   100
sample1   mir4   120
sample2   mir1   40
sample2   mir2   200
sample3   mir1   190

......

I need to change the directions of tthe above matrix, and an ideal output is like below.

          sample1    sample2    sample3
mir1      30           40         190
mir2      0            200         0
mir3      190          0           0
mir4      120          0           0
......

i tried to write perl hash of hash to sort out the problem (see below). However, I am new with this relatively complex hash. Could anyone provide suggestions. I believe it's a good stuff to learn for other perl beginners too. Thank you very much!

open FH, '<', $ARGV[0] or die "open failed:$!";
my %h;
while (<>){
        my ($sample, $mir, $abun) = /(.+?)\t(.+)\t(.+)/;
        $h{$sample}{$mir} = $abun; 
}
foreach my $sample (keys %h){
        foreach my $mir (keys %{h{$sample}})
                print "   "      # i am stuck here. Need your help!
}

[ericmajinglong]

May I ask, is perl an absolute requirement? What you are looking for can be accomplished by a few commands using Pandas in Python, and I am happy to provide code for you if you would like to.

[Irsan]

I agree with eric. Though it can be done with perl, it can more easily be accomplished by other means. In my experience, knowing how to work with R is very useful for numerical analyses. Have a look into the aggregate() function.

[biolab]

Thanks Eric and Irsan, I agree. Learning R is really necessary.

Answer 1

[Neilfws]

I realise that you asked for a Perl solution but as others noted in the comments, sometimes it's good to know that the right tool for the job exists already.

With that in mind, here is acast() from the R/reshape2 package.

library(reshape2)
mi <- read.table("mi.txt", header = T, stringsAsFactors = F)
acast(mi, MIR ~ SAMPLE)

#      sample1 sample2 sample3
# mir1      30      40     190
# mir2      NA     200      NA
# mir3     100      NA      NA
# mir4     120      NA      NA

[biolab]

Thanks a lot for all solutions. They are really helpful!

Answer 2

[umer.zeeshan.ijaz]

Okay quickly wrote this perl one-liner for you assuming that your input file is tab-delimited (just redirect the output to another file):

Best Wishes, Umer

[Devon Ryan]

This rather pushes the definition of "one liner" :)

[umer.zeeshan.ijaz]

saves u from one "wget" though ;)

[Alex Reynolds]

This answer doesn't make sense.

[Devon Ryan]

Looks like something got lost in the transition to the new biostars.