Scalability of methods
Here we test how well a method scales with increasing number of features (genes) and/or cells.
Each method is run on down- and upscaled datasets with increasing gene set and cell set sizes, and the execution times and memory usages are modelled using thin plate regressions splines within a generalised additivate model.
||Generate up- and downscaled datasets|
||Run the methods on the cluster|
||Retrieve the results and generate the scalability models|
||Classify the models and generate scalability figures|
||Helper to generate an up- and downscaled dataset which looks similar to the original datasets|
The results of this experiment are available here.