At some point you will need to get the data into the "usual" format with genes as columns (I am assuming you will be using those as inputs or predictors) and observations as rows. If you "transpose" the data to switch the rows and columns you can then merge the datasets using a Merge node. So you might try the Transpose node (under Field Ops) to see if that does what you need. You might test it out on a small subset of the data first.