I’m trying to run melonnpan.train on a filtered list of UniRef90 and metabolite features. Has anyone run into this problem before?
Error in { :
task 309 failed - “task 7 failed - “y is constant; gaussian glmnet fails at standardization step””
Hi @Chris_T - this will happen if, during the cross-validation, any of the response variables (i.e. metabolites) only contains a few non-unique values within a fold. Could you please make sure the metabolites have at least 5 or 10 unique values post-filtering as glmnet and melonnpan.train does 10-fold CV by default? Alternatively, you could also do a 3-fold or 5-fold CV if further filtering is not feasible. Hope this helps. Thanks!