Tool:microbiomeMarker: microbiome biomarker analysis in R

Question

[yiluheihei]

Hi, there

I'm developing a R package for microbiome marker discovery named microbiomeMarker, and the algorithm from lefse and stamp have been integrated to this package.

You can try it out today, if you want run lefse analysis in R. microbiomeMarker is still a newborn, so there may be bugs.

Thanks. Any suggestions and contribution will be highly appreciated.

[willnotburn]

I am running lefse using microbiomeMarker and it's taking forever (2 full days days so far). Is this expected? I am working with 140 samples with about 6000 species, and samples are in 5 clusters.

[Mensur Dlakic]

Maybe you should have left it at it's taking forever. When you add 2 full days days so far, it most definitely does not feel like forever, especially for 6000 species.

[willnotburn]

Oh, so you're saying this is normal? If yes, that's good news. There is no progress bar, and I have no way of knowing whether it's going to take 3 days or 30 days. Do you have any experience in this? How long is expected?

[Mensur Dlakic]

I don't know how long your analysis will take, but equating 2 days to forever struck me as a bit impatient. You may want to try it on a smaller number of species to make sure everything works, and potentially to read this discussion.

[willnotburn]

Thank you for the useful link. I will try at a higher taxonomic level.

The user is always right. Analyses that are expected to take a long time often have multithreading and memory cap options. They produce logs, usually displayed live to indicate progress. Documentation often warns of particular options known to take a long time (e.g. "use_heuristics=T (very slow)"). Without any of these features, the user of an interactive R session will start asking questions.

[MSRS]

Hi, thank you so much for the nice package.

I have an error in the package. It's not like an error.

I have a phyloseq object

phyloseq-class experiment-level object
otu_table()   OTU Table:         [ 11 taxa and 7 samples ]
sample_data() Sample Data:       [ 7 samples by 10 sample variables ]
tax_table()   Taxonomy Table:    [ 11 taxa by 7 taxonomic ranks ]

When I normalized the physeq object by using the following code,

ps <- norm_css(physeq, sl = 1000)

Unfortunately, there was no change in the OTU table, or there was no transformation that occured.

CSS, TMM, RLE not working here.

is there any suggession?