Track gap filling / imputed values

[wolski]

This use case comes from Metabolomics/Compound discoverer, but will also open the option for modeling imputed data.

• CD reports in the Gap Status column how compound abundance was generated (e.g., Gap Fill - imputation, I guess).
• This is reported per compound and sample.

The idea is to keep track of this information in the LFQData object,

This can be implemented as follows :

• in addition to the response, use a gap_fill_status / nr_children column to track whether the response is observed or imputed.

Implementation:

• use just introduced nr_children column - if Number children 0 => imputed.
• Currently, we compute statistics about observed and missing values using the complete_cases method and counting NAs.
  Options:
• If gap_fill_status column is present, use an alternative implementation. Otherwise, use existing.
• Or always present, then use nr_children_column and dichotomize on 0 and not 0.

For the moment, I will go with two versions.