DESCRIPTION ----------- The first GenoGAM package was a first step towards introducing generalized additive models (GAM) to the biological field. The implementation was mostly done around the excellent mgcv package, which provides a very flexible GAM framework. The flexibility however comes at a price of increased runtime and bigger overhead. With the new version of GenoGAM, mgcv was replaced by a GAM framework specifically optimized for biological applications, significantly descreasing runtime and overhead. NEW FEATURES IN VERSION 2.0 ----------------------------- o Simplified underlying infrastructure due to replacement of the mgcv package at the core of GenoGAM by an inhouse build GAM model optimized for biolocial applications. This significantly reduces overhead and thus runtime and memory consumption. o GenoGAMDataSet can now be efficiently created from SummarizedExperiment objects o Access to the GenoGAMSettings class, to modify global settings for computation of the model o Included logging on different levels for easier maintanance and debugging o Included second penalization (first order) for regions with very low coverage. o Included extensive unittests for all functions o HDF5 backend for large datasets such as the human genome FEATURES THAT ARE MISSING FROM VERSION 1.0 ------------------------------------------ Those features are currently missing but will be included with the next release o Quality check plots o Estimate of the peak width in peak calling o Filtering o ggplot2 plotting for GenoGAM fit objects o access to knots and coefficients of the model FEATURES IND DEVELOPMENT ------------------------------------------ Those features are currently being implemented (alongside with the ones above) o More distributions (in particular quasi-binomial and gaussian) o better quality check plots o logging to file and in parallel environments o plotting as Gviz objects