How to deal with missing continuous metadata

I’m wanting to analyze 10 metadata variables (8 continuous and 2 categorical) against metabolite data using the Statistical Analysis (metadata) function on MetaboAnalyst. However, I have a significant amount of missing values in my continuous metadata variables, of which I am unable to assign to “unknown” or “NA” categories as MetaboAnalyst requires the value to be numerical if the data is continuous. I don’t want to input the value as 0 as this will manipulate the data, nor do I want to delete whole samples due to them missing one value of a continuous variable.

Is there a way around this missing continuous data without having to delete entire samples or creating new spreadsheets with each metadata variable on?

We have the same concerns as you mentioned. I can think some possible solutions:

  1. Dynamically slice/subset the data to contain non-missing values for this metadata variable (assuming your data is relatively large) - this data slice will be used if you are interested in this variable;
  2. Impute missing values based on remaining data, with cautions of potential bias (here is a paper on this topic)

These features are not available at the moment. You may try to use some R functions to test out.