These are executable files for computing normalization and a brief description of how they are implmented:

• find_ML_normfactors3D: calls ML_estimate_component_based_normalisation and the single function there. Additionally, the documentation in ML_estimate_component_based_normalisation.h is lacking.
• find_ML_normfactors: implements much of the functionality in the main function.
• apply_normfactors3D: implements things in the main function, at least is very short.
• apply_normfactors: does the same as apply_normfactors3D (code duplication?)

Most of the following pertains to find_ML_normfactors3D and find_ML_normfactors. The above functionality saves the computed data to be saved to disk, rather than returning/utilizing a shared_ptr.

The core functionality of these executables should be split into various components.

1. The main function should handle only the arguments handling and saving data.
2. Mostly, functions/classes that take the core arguments and compute/process the data.
3. Can handle saving of the data, if specified by the arguments.
4. Function/ object returns a pointer to the computed data, that can be saved by the main exe.

To implement this:

• The exe's need to maintain backwards compatibility.
• Replace the ML_estimate_component_based_normalisation function with a class/function..
• Allow member variables/argument to determine if data is saved to disk.
• Store/return computed data as members that can be saved/handled by the exe main functions.