Variable Type

[alina_...@brown.edu]

Hi, 

I have a question regarding the interpretation of variable types by Maaslin. I read online that "the clinical metadata can be of any type continuous (for example age and weight), boolean (sex, stool/biopsy), or discrete/factor (cohort groupings and phenotypes). Does this mean Maaslin is unable to find associations with categorical variables?

For example, if I have a metadata variable X where 

1 = 1-10 years old 

2 = 11-20 years old 

3 = 21-30 years old 

4 = 31+ years old 

Would Maaslin treat variable X as an ordinal continuous type variable? If not, how would Maaslin generate associations with metadata variable X and the species abundance data? 

Thank you, 

Alina Husain

[Jeremy Wilkinson]

Hi Alina,

Variable X in your dataset will be treated as a continuous variable as is: 1, 2, 3, 4. If you want it to be treated as a discrete variable, then you can add a character to the beginning or end of the level such as an underscore to change the variable from numeric to character. For example: 1_, 2_, 3_, 4_. 

-Jeremy