This is a test version of Biostars. For the public version, visit https://www.biostars.org.
LEfSe for multiple samples

I have 5 groups of samples, all with different abundances of various genes (KOs). Is there a LEfSe implementation that compares 5 groups at the same time?

R Bioconductor has a user-friendsly package lefser, but it only takes two groups at a time as input.

rna-seq metagenomics environmental genomics
• 3,967 views • link
updated by h.mon h.mon • written by willnotburn • willnotburn

You can use lefser for that propose. Have a look into the argument one.against.one : for multiclass tasks, sets whether testing is performed one-against-one (TRUE - more strict) or one-against-all (FALSE - less strict)

So if you have 5 groups: A,B,C,D and E, for the all-against-all feature has to be differential abundant between all the 5 groups (feature p is detected if its abundance is different in A vs B, A vs C, A vs D, B vs C...) . And for one against all: Feature p is detected if its abundance is different in A vs B, A vs C, A vs D but not necessary in A vs E for example)

• 0 views • link
written by Lila M Lila M

That would be exactly what I need! But I'm not seeing those arguments in lefser. Is there a development version? I installed mine off of Bioconductor, and it's Version: 1.0.0

lefser(
  expr,
  kruskal.threshold = 0.05,
  wilcox.threshold = 0.05,
  lda.threshold = 2,
  groupCol = "GROUP",
  blockCol = NULL,
  assay = 1L,
  trim.names = FALSE
)
• 0 views • link
written by willnotburn • willnotburn

you can try this. Other option is to use Galaxy m that has a very good implemented interface to address this issue. Is less manipulative but is very straight forward

• 0 views • link
written by Lila M Lila M

Thank you! The lefse function in yingtools2 seems like a more mature implementation than a standalone lefser package.

• 0 views • link
written by willnotburn • willnotburn

0 answers

No answers yet.

Log in to answer this question.

More posts like this