Shell Programming and Scripting

Code:

nawk '{s=0;for(i=1;i<=NF;i++) s+=($NF-$i)^2;print sqrt(s)}' slice.txt

Thanks,
guys this solved the issue.

Am very grateful for your patience, and your support.

X

---------- Post updated 01-13-12 at 05:58 AM ---------- Previous update was 01-12-12 at 04:37 PM ----------

Hi guys,
a little extra question,

what if I wanted the program to work in the reverse manner,

parse ONLY two columns per iteration, e.g. column 1 ($1) and column 50 ($50) and subtract value 1 in column 1 from value 1 in column 50, etc. until I calculate one RMSD number.

THEN move onto column 2 ($2) and column 50 ($50) etc. and obtain a second RMSD number.
etc etc until it is column 50 ($50) vs. column 50 ($50).

?

Currently my script parses each line separately, and it is great, but I want to try the other way around.

---------- Post updated at 06:09 AM ---------- Previous update was at 05:58 AM ----------









so instead of NF I thought of using NR to go down the column instead of across?

---------- Post updated at 06:45 AM ---------- Previous update was at 06:09 AM ----------

So for instance,

If I wanted to obtain the RMSD between column 1 and column 50 I wrote

Code:

{s=0;NF==1;for(i=1;i<=NR;i++)} s+=($NF-$50)^2;print sqrt(s/NR)}

but I get an error:

Code:

awk -f rmsd4 merge.pmf > test2
awk: syntax error at source line 2 source file rmsd4
 context is
    {s=0;NF==1;for(i=1;i<=NR;i++)} >>>  s+=($NR-$36)^2;print <<<  sqrt(s/NR)}
    extra }
awk: bailing out at source line 4

If you could post a sample of your data, a sample of what output you want, and show your calculations that would be much, much better than trying to explain an algorithm in casual English. It's a little more work for you, I understand, but it also means a much greater chance of being understood on the first try. And being all you need is a single line of data to demonstrate, it really isn't so awful. It may also help you get it organized better and see an algorithm yourself.

I don't understand what you're getting at now, and since your code doesn't work, it's not a good demonstration either.

I think I see your syntax error -- an extra bracket:

Code:

{s=0;NF==1;for(i=1;i<=NR;i++)} s+=($NF-$50)^2;print sqrt(s/NR)}

There's logic errors too, though.

The statement in blue, though, what did you intend that to do? Right now it's a complete no-op.

NR is the number of lines(records), not the number of fields, I think you want (i=1;i<=NF;i++) and sqrt(s/NF)

I also think that's off by one, since that will include the last column, the average itself, so:

(i=1;i<NF;i++) and sqrt(s/(NF-1))

And then, that thing in green. Since NF doesn't change until the number of columns does, this is always adding the same columns: ($NF-$50) In fact if you have 50 columns, $NF will be column 50, causing the result to always be zero!

I think the OP wants to do the calculations based on columns, not based on rows as initially stated. But I fail to understand the algorithm entirely - using 'NR' instead of 'NF' is not going to achieve the desired results.
One would need to parse all lines and build up a 2-D matrix and navigate it by columns performing the desired calculation.
Once again, this is all somewhat nebulous for me.
If the OP could demonstrate what he/she/it is after given a simple table (say 3x3 - not necessarily Nx50)

Code:

1 2 3
4 5 6
7 8 9

and a sample output - that would be helpful.
[/code]

Hi,
ok,
I understand I was not clear enough.

Code:

BEGIN {s_0=0;n_0=0}
      {n_0++;s_0+=($50-$1)^2}
END {print sqrt(s_0/n_0)}

BEGIN {s_1=0;n_1=0}
      {n_1++;s_1+=($50-$2)^2}
END {print sqrt(s_1/n_1)}
...
BEGIN {s_50=0;n_50=0}
      {n_50++;s_50+=($50-$50)^2}
END {print sqrt(s_50/n_50)}

I have written a code to work on a document with 50 columns of data to calculate the RMSD of each column of data relative to the last column of data.

The script is pretty primitive, and involves 50 different variables. Is there a way I could make it automatic by simply incrementing the n_x variable instead of having to define the 50 variables manually. Also, instead of defining the last column as $50 how do I do it with an $NF?

How about that sample input and output?