Help with LEfSE

Wellinton Muniz

unread,

Aug 28, 2019, 3:47:04 PM8/28/19

to LEfSe-users

Dear All,

I have a trouble with the LEfSE analysis. Briefly, i obtained the file in txt. I have tried to modified this file adding two raws in order to include Class- Subclass but when I switch to Galaxy there is this problem:

Traceback (most recent call last):
File “/shed_tools/testtoolshed.g2.bx.psu.edu/repos/george-weingart/lefse/a6284ef17bf3/lefse/format_input.py”, line 435, in
feats = numerical_values(feats,params[‘norm_v’])
File "/shed_tools/testtoolshed
31 A) Format Dat

I thought that y file was correctly formatted but I do not understand the problem or how to provide a correct txt.

Thanks a lot for your help

Jérémie Auger

unread,

Aug 29, 2019, 11:05:10 AM8/29/19

to LEfSe-users

Are you running lefse locally (with installed version on your PC)? You can instead try :

http://huttenhower.sph.harvard.edu/galaxy/

I find that the GUI actually helps! And you will be sure that your data is correctly formatted. Lefse is very picky (like is the file's extension is *.txt instead of *.tsv, it will not even open the file and reply 'upload a correctly formatted file'..)

I don't want to download random files, but if you could upload the head (first like 10 lines) from your table, I'd be able to help better.

Cheers, -JA

Jérémie Auger

unread,

Aug 29, 2019, 1:30:46 PM8/29/19

to LEfSe-users

Don't use a "excel" file format. Copy your whole table and paste to a text raw editor, I guess you can use notepad (basic from windows). I would suggest installing NotePad++, which is a great free tool (a definite must even) or Atom (which is cross-platform and also free, developed by github)).

Then, save this raw text file with an extension "*.tsv"

Jérémie Auger

unread,

Aug 29, 2019, 1:58:20 PM8/29/19

to LEfSe-users

And also! Your 'ID' line (in the header of the file, second line). In lefse, only add the sampleID if they are repeated measure. I look, and all the IDs are unique numbers. Remove the line, doesn't serve! (only keep data that can help make groups)

Also, Lefse is nice on data that splits onto sub-grous (ex. the proteobacteria splits in Alpha, beta, gamma-proteobacteria). These tables seem te be single taxonomic level.

Jérémie Auger

unread,

Aug 29, 2019, 2:07:42 PM8/29/19

to LEfSe-users

And also! Your 'ID' line (in the header of the file, the second line). In lefse, only add the sampleID if they are repeated measure (from participant #17 at visit V1, V2 and V3). I looked, and all the IDs are unique. Remove the line, it doesn't serve! (only keep data that can help make groups)

Also, Lefse is nice on data that splits onto sub-grous (ex. the proteobacteria splits in Alpha, beta, gamma-proteobacteria). These tables seem te be single taxonomic level and won't have that multi-layer output.

Bonne journéee, -JA

Jérémie Auger

unread,

Sep 6, 2019, 1:23:20 PM9/6/19

to LEfSe-users

So, did you get it working? If not, I have a bit of spare time today, send table, I'll format it for you!

Also, I noticed that there is a hierarchical structure in the few names you sent me. Make sure to use the pipe character to divide the ranks (the pipe is the vertical bar | )