Re: calculate average in sliding windows

[marfabe]

thanks to all...but I still have problems in the sliding window part
Maybe I didn't explain it ok. What I want is to group the first column of the file in windows of 10 kbases (DNA positions) and calculate the average of column 2 for each window.

BrowserUK and Eliya: your idea works fine but it doesn't take into account the positions in column 1...and I'm stuck trying to use it.
Any idea?
Thanks!

[Eliya]

but it doesn't take into account the positions in column 1

Sorry, I'm still not sure what you want (and it's a little diffficult to infer it from code which - as you state - doesn't work).

Are you saying the positions in column 1 are meant to represent indices in some sparse array, with all unspecified positions (such as 12498249..12512574, etc.) implicitly having a value of zero, or what?

I think it would help if you could come up with some simplified (but sufficient/complete) sample input (say using a window size of 3) together with the desired output values.

[marfabe]

Thanks Eliya
Here's the simplified sample set


position     value
1    1
10    3
30    1
40    2
60    2

And here the output for window size of 30


1-30   1.666666667
31-60  2

[Eliya]

Ok, thanks.  So maybe like this?

#!/usr/bin/perl -lw
use strict;
use constant WINDOW => 30;

my $sum = 0;
my $n   = 0;
my $p   = WINDOW;

while (<DATA>) {
    my ($pos, $val) = split;
    if (eof) {  # corner case
        $sum += $val;
        $n++;
    }
    if ($pos > $p or eof) {
        print $sum / $n if $n > 0;
        $sum = 0;
        $n = 0;
        $p += WINDOW while $pos > $p;
    }
    $sum += $val;
    $n++;
}

__DATA__
1    1
10    3
30    1
40    2
60    2

Output:

1.66666666666667
2

(Upd: fixed handling of corner case)