Hello Joachim,
In short, you would want to avoid mixing rows of containing a different number of columns or empty values being used in the calculation. So, this would be possible as long as:
1. There are an equal number of columns per row in the file overall. This may mean that "null" or empty values have to be padded.
2. Any null or padded values are either not included in the calculation or are padded with a "0", if that still makes sense for your calculation.
If this doesn't seem to meet you needs, perhaps filtering the data into different populations, performing the calculation, then merging the result is an option (as would grouping the steps into a workflow). Using tools such as "Filter" or "Select" with "Cut" and "Paste" would likely be helpful.
Good luck!
Jen Galaxy team
On 12/10/12 5:20 AM, Joachim Jacob wrote:
Hi all,
I wonder if there is a method for calculating a metric (e.g. sum) per row (working on a tabular dataset). But, each row contains a different number of values (i.e. some columns are empty). Can I compute this, using Compute?
Cheers, Joachim