get_normalized wrapper function

[drisso]

To give some more details, the reason we may need this is because in scone some of the parameters actually specify a range of values in the function, e.g., k_ruv=5 actually means k_ruv=1:5.

Perhaps, we should change the behavior of these parameters rather than creating a new function.

[drisso]

This doesn't seem like a straightforward fix. Need some discussion with @mbcole to decide how to proceed. I'm moving it out of the milestone.

[mbcole]

Agreed - if we change the argument defs then this won't be necessary.

[drisso]

I now think that a get_normalized wrapper would actually be straightforward.

We need a wrapper around this (loosely):

• input: a scone object that has already the evaluation + the string corresponding to the desired norm
• run scone with run=TRUE, eval=FALSE after grepping the input string from the params data.frame
• return only the normalized data part of the scone output

For this to be useful, we need to reintroduce the possibility for the user to run the evaluation without saving the normalized data. This is implemented with the return_norm argument in the feature branch wrapper (commit 4f8e761837ee5e98ddd97e11da16fd2a9e23e6eb).

[drisso]

I've created a get_normalized wrapper that works if the user saved the results to file.

We should eventually add the possibility to use it also if the normalization was not saved as it can be computed on the fly. This means a fair amount of coding, which I don't really want to do now, because it would require a lot of re-writing when we switch to S4. S4 will naturally solve this problem by keeping all we need in the same object.