Hello,

I recently installed Galaxy and among other tools I would like to have LEfSe on my local instance.
I get the latest LEfSe's repository from the galaxy main tool shed, and I also install it's requirements packages and R libraries on the server (btw it's Debian based).

I have a problem with the first tool, "A) Format Data for LEfSe", as you can see in the picture I attached "galaxy_lefse_error.png". In this step, I'm supposed to indicate which row Galaxy has to use to label the data. My issue is that the box "Select which row to use as class" remains saying "No options available", which blocks me to process the step. After checking the xml file of the tool, I found that a "dynamic option" is involved and normally have to generate the options list.

I got helped by George Weingart (thanks to him and his patience) and I can say the following things :
- My file is correct : I can run LEfSe on it with the public server Huttenhower-lab. I did the same manipulations with the example file provided by LEfSe's authors (found at this link http://www.huttenhower.org/webfm_send/73), and same result on my own Galaxy instance, while it's working on Huttenhower-lab;
- This issue is not related to LEfSe : I installed the program outside of Galaxy (as stand alone) on the server and I get correct results. I tried to uninstall/reinstall LEfSe in Galaxy from the tool shed. I also replaced the LEfSe's scripts in the tool shed folder by the ones of my stand alone LEfSe. I checked the Galaxy log and didn't find any error possibly related to LEfSe. And finally, the other tools of LEfSe in my Galaxy instance are working well and give me correct results, only the "format data" tool has a problem;
- With the same file (e.g. the example one), a difference occurs in the data loading between my Galaxy instance and Huttenhower-lab. After uploading, the datatype detected by my Galaxy instance is "axes", whereas it's "tabular" with Huttenhower-lab. Manually forcing the datatype to tabular (on upload or after upload) doesn't solve my issue.

Does anyone have a clue to explain this difference in datatype detection?

Many thanks in advance,
Romain