metricView: Internal function called by 'runDVboostwrapper'

In Liuy12/DVboost: Adaptive Variant Filtering using generalized boosted regression models

Internal function called by runDVboostwrapper

Description

After model training and testing, metricView generates summary statistics and figures for users to examine the training performance

Usage

metricView(DV.fit.res, input.sample.ID, plot.filename,
  DV.sampleQC.filename, subset)

Arguments

DV.fit.res	output generated by fitDVboostmodel
input.sample.ID	same sample ID used for annotation procedure
plot.filename	full path to the figure file regarding training performance
DV.sampleQC.filename	full path to the text file that contains metrics for model training performance
subset	a vector of integers indicating a subset of the SVs that all metrics will be evaluated on

Value

A figure file

The figure is saved at plot.filename and contains 6 subpanels, including

1. distribution of known(1) versus novel(0) variants;

2. feature importance for model training;

3. distribution of Q scores for known and novel variants;

4. precision, recell characteristics and best F score;

5. ROC curve;

6. FDR versus Q score

QC metrics file

The .txt file is saved at DV.sampleQC.filename and contains 6 fields, including

• sample.ID: input sample ID

• Fscore_PR: F score for precision&recall

• AUC.PR.known.var: Area under the curve for precision and recall

• AUC.FP.TP.known.var:AUROC

• npos:number of known variants used for training

• nneg:number of novel variants used for training

Liuy12/DVboost documentation built on May 25, 2022, 6:17 a.m.