bug? Negative values treated as missing values

[eschen42]

Negative values are treated as missing values. This makes the tool not functional for processing residuals. Is this a bug? If not, why would it make sense?

expected: default values (no filtering) should leave the data matrix unchanged.
actual: negative values are replaced with imputed values or NA when imputation is off.

example:
input_zerobug.zip

[eschen42]

This would make sense for log-transformation of data, but is there a need when not log-transforming data?