I realized LEfSe provides different results from identical input files in which only feature names have been modified. Compare the outputs below: the number of significant feature differ by 1 and the LDA scores changed for some features.
Anyone faced the same issue?