Warning: More intervals than unique values.

[meret-haeusler]

Hi everyone,

I am currently working on a project where we want to apply R.ROSETTA to a metagenomic dataset (28 samples/objects and 121 taxa/features).

For both, the Johnson and genetic reducer, I get a lot of warnings that look like this:
WARNING: More intervals than unique values.
WARNING: No cuts found for attribute x.

Unfortunately, I could not find any explanations for these warnings in the documentation.
I think that my data set has too few objects for R.ROSETTA to work properly. Is this correct or is there another explanation?

Thanks in advance!
~Meret

[mategarb]

Hi Meret,

These warnings come from discretization step. It says that your data set may be categorical for some reason. Is your dataset discrete (categorical) already?
If yes, you need to set 'discrete' option in R.ROSETTA to TRUE. Here is tutorial, how to do it:
https://komorowskilab.github.io/R.ROSETTA/tutorials.html#discrete

Unforunately, there is no documentation for warnings as they come from the ROSETTA tool that is inside the package (it is a wraper of command line ROSETTA version). There is a manual for ROSETTA http://bioinf.icm.uu.se/rosetta/materials/manual.pdf that may additionally clarify some parameters and functions.

I think your data should have enough number of objects, also it depends how are they distributed for decision classes.

If there is anything else, please let me know.

best wishes,
Mateusz

[roymey]

Hi Sir,
Can you please help me with how to apply a dynamic reducer? From the manual, I did guess that this code will work using reducer="RSEDynamicReducer" will work. But unfortunately, it does not!

[mategarb]

Dear Roymey,
In the R version of ROSETTA, RSE-based functions were not implemented. The available reducers are Johnson and Genetic.
best,
Mateusz

[roymey]

thank you :)