Problem to use mofaOBJECT

[geraldocantelli]

Good afternoon,
I am trying to use MOFA with my own data but when I create a MOFA object and try to set the environment to run MOFA I get the following error;

$scaleViews
[1] FALSE

$removeIncompleteSamples
[1] FALSE

Error in data[!is.na(data)]%%1 :
argument non-numéric to binary operator
Calls: getDefaultModelOptions -> .inferLikelihoods
interrupts

My code is:
library(MultiAssayExperiment)
library(MOFA)
library(MOFAdata)
library(reticulate)

Using a specific python binary

use_python("/usr/local/bin/python")

transcriptome <- read.table ("/home/geraldo/eclipse-workspace/wsapp/WebContent/apprMOFA/transcriptome.csv", sep = "\t", header = FALSE)

dfT <- data.frame(transcriptome)

metabolome <- read.table ("/home/geraldo/eclipse-workspace/wsapp/WebContent/apprMOFA/metabolome.csv", sep = "\t", header = FALSE)

dfM <- data.frame(metabolome)

allLists <- list ("transcriptoma" = dfT,"metaboloma" = dfM)

#data("CLL_data")

MOFAobject <- createMOFAobject(allLists)

DataOptions <- getDefaultDataOptions()

DataOptions

ModelOptions <- getDefaultModelOptions(MOFAobject)

TrainOptions <- getDefaultTrainOptions()

TrainOptions

MOFAobject <- prepareMOFA(
MOFAobject,
DataOptions = DataOptions,
ModelOptions = ModelOptions,
TrainOptions = TrainOptions
)

outfile = file.path(getwd(),"model.hdf5")

MOFAobject.trained <- runMOFA(MOFAobject, outfile)

Could someone say how to set inferLikelihoods correctly or what can I do in order to code run, please?

The file data: metabolome.csv
metaboloma spontL1 sp80
metaboloma A 10.0 1.0
metaboloma B 5.0 5.0
metaboloma C 1.0 10.0
metaboloma A 10.0 1.0
metaboloma B 5.0 5.0
metaboloma C 1.0 10.0
metaboloma A 10.0 1.0
metaboloma B 5.0 5.0
metaboloma C 1.0 10.0
metaboloma A 10.0 1.0
metaboloma B 5.0 5.0
metaboloma C 1.0 10.0

And transcriptome.csv is the same with the word "transcriptoma".

Best regards,
Geraldo

[Ricard Argelaguet]

Hi Geraldo,
is the data normalised in any way? the input data should be continuous, not discrete.

P.S. Please switch to MOFA2 (https://biofam.github.io/MOFA2/), which is also available via bioconductor (http://bioconductor.org/packages/release/bioc/html/MOFA2.html). MOFA v1 is depreciated

[geraldocantelli]

Thank you very much for the answer.
So I was testing with wrong values, I am going to use continuous data now and I will update to MOFA2.
Best regards,
Geraldo