In gstricker/fastGenoGAM: A GAM based framework for analysis of ChIP-Seq data

To implement

Quality checks. Just like logging on different levels
Logging has to be output to files if threshold is debug
Plotting
differential binding
peak calling
export function
suppressWarnings does not work in tile construction if parallel backend is Snow
make exportable cross validation function
include gaussian
include quasi-binomial
get rid of warning messages during read data on makeTile level and on readingBAM level
deactivate ncv control within genogam
make predict function
keep in ggs object the design "i,e eg (1,0,1,1)" and the spline templates matrices "ie, eg (X0, X1)". when computng spline errors i) do not rextract those and ii) compute directly the spline standard errors just once using and then put them back into the returned object by replicating values. You will gain a factor 4 on experiment with duplicates and much more if we apply it for large datasets (say groups of 100 vs 100 patients).
add logging in parallel computation (see vignette on BiocParallel: Logging)
set #define ARMA_64BIT_WORD or check if it is set and throw warning that matrices should not exceed certain size.
Minimal Lambda inclusion