Lefse input file: how to set the experimental design

Hi!
I am trying to run LefSe analysis in Galaxy and I am having problems with the input file. I found the make.lefse tool in Galaxy and it requires the .shared, .constaxonomy and the experimental design file; I’m not sure what the experimental design should be like, anybody could give me an example so that I can adjust it to my own data?

I would suggest to make your own feature/otu/asv table outside of galaxy before starting Lefse analysis. In which case, you would have your OTUs as rows in the first column, and the samples for your class and subclass (if appropriate) as your first and second rows.

It is essentially a transposed shared file with the label column removed.

Let me know if you require more clarification.

I am working on microbiome analysis of lung cancer patients and want to perform lefse my design file input is confusing, the other two inputs i have already. how should the design be like so that it is acceptable. any visual demonstration of it? or if someone could guide me. regards

Hello,
My best suggestion would be to look at the sample data shown on Galaxy, under the tab called A) Format Data for LEfSe . Partway down the page there is a header called “Input File Format” that gives some guidance.
I hope that is helpful!
Best,
Meg

1 Like


https://huttenhower.sph.harvard.edu/galaxy/datasets/2419d42db6e7bdd7/display/?preview=True
the above image is the output of the make.lefse file used as input in galaxy hutlab. the output does not mention SCLC samples as were mentioned in the input. is there any solid reason for this? if someone could guide me. also, i have three different components of my data for analysis. how can i merge three make.lefse files into one file for visualization? i am concerned about these two problems. thanks in advance.

Thank you so much for your pertinent response. May good come to you.

Hello,
I think you might have a typo in column 9, where the category is listed as NSClC instead of NSCLC, and so lefse is treating this one sample as its own category. Try changing the group label for that column and re-running, hopefully your answers will make more sense. As for the same visualization, I’m not sure exactly what you’re asking–are you looking to combine the output plots from an analysis into one figure? If so, it’s not something that LEfSe can do, you’d need to use a software such as Adobe Illustrator, inkscape, or Powerpoint.
I hope that helps!
Meg