# Backlog / issues

## Pipeline
* `bbAlign`: Determine where "No_fill" peak table should be taken from.
* Before hard fill or after hard fill/before imputation?
* How to evaluate? Somehow by checking NA-values...
* _Date added_: 2024-06-04 (AR)

* `driftCorr`: Scaling of variables, which order to do things:
* After within-batch drift correction
* Anton/Calle pseudocode to manage zeros and missingness (see issue with "Problems with 0z in dataset")
* _Date added_: 2024-06-04 (AR)

* Problems with 0z in COSMC dataset
* Arises from imputation - mvImpWrap bug fixed 2024-06-04 (CB)
* What to do?
* Set `pp_imp[pp_imp < 0 ] <- 0`
* perform `bbAlign`
* Per batch
* `batchCorr`
* imputeSubZero from batchCorr with `mvImpWrap`
* Feats which are 100% 0s are flagged
* rnorm() to give a spread for normalization not to break
* scale from zero (CB: I don't understand what this means)
* merge batches
* normalize batches
* remove flagged values (entire features within batches)
* impute missing values (i.e. systematic imputation of en entire feature within specific batch(es))
* _Date added_: 2024-06-04 (AR)

* IPO2 rtAlign optimization
* None-existent, there is no way to process using xcms4 while this problem persists
* _Date added_: 2024-06-06 (AR)

* IPO2 PP optimization
* Should be xcms4 according to Calle
* _Date added_: 2024-06-06 (AR)

* Parallelization of plotClust
* One pdf takes 6 hours to generate in COSMC
* Faster plotting + parallelization is needed
* _Date added_: 2024-06-06 (AR)

# Fixed issues

* `mvImpWrap()`: Gives negative values, even with `forceZero = TRUE`
* Isolate bug: Second round mvImp in mvImpWrap was missing the forceZero argument
* Fix: Added forceZero argument to 2nd round mvImp in mvImpWrap.
* Additionally: Added some more output and cleaned up some code
* _Date added_: 2024-06-04 (CB)
* _Date fixed_: 2024-06-04 (CB)