Potential m3n issue
Issue #14
resolved
histogram for distToNearest with m3n has a significantly tinier second mode than any of the other four methods. Possible bug in targeting model, may need a double-check.
Comments (3)
-
-
- marked as critical
-
- changed status to resolved
We removed the M3N model entirely.
- Log in to comment
The 3-mer mutability is missing values (ie they are mutability = 0 in the Smith et al, 1996 paper), which appears to be the source of zeros and extra NAs in the targeting model. Also, the mutability values in
data-raw/M3N_Mutability.csv
don't match those in the Smith et al, 1996 paper.This might not be the only problem, but it is a problem... I think we need to impute the missing data.