# Multiple combination of column aggregation

Greetings to all,

I have a matrix of float values where I would like to make all possible combinations (mostly up to 5 columns combined) of columns through an aggregator node, where the output for each row would be the minimal value of the column values combined.

Would it be possible to make such output table?

Practically, the name of the table would be the name of two (or more) columns combined, let's say column a and column be would become column a_b and the values for each row of that column would be the minimal value of the combined columns for that row.

Hi Filip,

However doing all combinations will always be pretty ressource intensive.

Cheers, Iris

Hi Iris,

Thanks for the sent workflow, but, maybe I have expressed myself a bit wrongly...

Imagine I had 3x3 matrix, in each cell there would be double (float values). I take first two columns, compare first row for the values, take the minimum, write it for the same row in another column, then go to the next row and then repeat the process:

 Ligand RMSD_4l6q RMSD_4wot RMSD_5u7r 4L6Q.A_entry_00001_conf_01 1.53 8.1 0.49 4WOT.A_entry_00002_conf_01 2.64 0.82 2.2 5U7R.A_entry_00003_conf_01 2.04 5.43 0.82

If we combine RMSD_4l6q and RMSD_4wot and take the minimum value for each row of their combination, the output would be:

 Ligand RMSD_4l6q_RMSD_4wot 4L6Q.A_entry_00001_conf_01 1.53 4WOT.A_entry_00002_conf_01 0,82 5U7R.A_entry_00003_conf_01 2.04

I want to do practicaly this for all combinations of columns and then see which combination gives the best result.

Would that be possible? Maybe it requires some looping but I am not sure how to achieve that.

Thanks

Hi again Iris,

I think that I have managed it, thanks to your posted solution which gave me an insight how to deal with the combinations part. If you want, I can post the solution, even though it would need some cleaning.

Cheers!

Filip