enhancer_classifier: Enhancer classifier based on centrality measures and...

In LucaGiudice/Esearch3D: Leveraging gene expression and chromatin architecture to predict intergenic enhancers

Description

This function creates a classifier able to predict if a node is an enhancer or not based on centrality measures and propagation values obtained with the multi-gene two-step propagation. First, it creates the task, tunes a ranger random forest, trains the model and predict a randomly selected testing set in cross validation.

Usage

enhancer_classifier(
  info,
  proj_name = "enh_pred",
  annot = 2,
  n_cores = 2,
  perf_measure = "classif.acc",
  tune_perf = 0.7,
  n_folds = 5
)

Arguments

info	Info dataframe containing centrality measures and propagation values about the fragments and their corresponding nodes
proj_name	name fo the project as character string
annot	Define the minimum number of annotations to define an enhancer (e.g. 5)
n_cores	Define number of cores to use in order to parallelize the tune, train and test tasks
perf_measure	Set performace measure (e.g. classif.acc)
tune_perf	Performance to reach in the tuning phase (e.g 0.7 of accuracy)
n_folds	Number of folds for the cross validation (e.g. 5)

Value

List of 7 elements: proj_name, perfs_df, data_l, task, instance, lrn_tuned, lrn_tested, best_lrn proj_name is the name of the project and run perfs_df is the dataframe containing the classification performances data_l is the list containing the original data processed by the classifier task is the mlr3 object containing the task of classification instance is the mlr3 object containing the tune, train and test parameters lrn_tuned is the mlr3 object containing the classifier after the tuning lrn_tested is the mlr3 object containing the classifier after the testing best_lrn is the mlr3 object containing the classifier which best classifiers enhancers

LucaGiudice/Esearch3D documentation built on Dec. 17, 2021, 1:12 a.m.