Hi,
I am new to LEfSe and have a basic question about the results provided by LEfSe. The tool provides 5 column table with no column heading. I understand that, column 1 is the feature being tested.
I am also wondering the same thing.
Hello,
The documentation online describes the first four columns of output:
“The output consists of a tabular file listing all the features, the logarithm value of the highest mean among all the classes, and if the feature is discriminative, the class with the highest mean and the logarithmic LDA score.”
For some reason the fifth column is not described, but it contains the p-value from the kruskall-wallis test for differences in the feature among the class variables, given that the K-W test p-value is below the specified threshold (default is p<0.05). This does not include a correction for multiple testing. The K-W test is just the first step in identifying potentially important features; effect size measurement and assessing effects within subclasses also helps to narrow features to those more likely to be related to the class variable of interest. Setting the p-value threshold to be very low (such as with a Bonferroni correction for the number of features) is possible, though not necessarily recommended. The paper contains some thoughts about this: https://doi.org/10.1186/gb-2011-12-6-r60
I am working on the differential abundance in microbiome visualisation using lefse
Do I need to correct the p-value while using lefse? when I have two groups to compare?
Hello,
The use of LDA score filtering in addition to significance filtering is used to identify features that are most likely to be associated with the class variable. There is some amount of personal preference around whether you consider this to be stringent enough for your needs–see this conversation thread, which gets into this a bit.
I hope that helps!
Best,
Meg