Skip to contents

Compute evaluation report for TMB/DESeq2 analysis

evaluation_report.Rd

This function computes an evaluation report for TMB/DESeq2 analysis using several graphical summaries like precision-recall (PR) curve, Receiver operating characteristic (ROC) curve and others. It takes as input several parameters like TMB results (l_tmb), DESeq2 result (dds), mock object (mock_obj), coefficient threshold (coeff_threshold) and alternative hypothesis (alt_hypothesis).

Usage

evaluation_report(
  mock_obj,
  list_gene = NULL,
  list_tmb = NULL,
  dds = NULL,
  coeff_threshold = 0.69,
  alt_hypothesis = "greaterAbs",
  alpha_risk = 0.05,
  palette_color = c(DESeq2 = "#500472", HTRfit = "#79cbb8"),
  palette_shape = c(DESeq2 = 17, HTRfit = 19),
  skip_eval_intercept = TRUE,
  use_initial_random_params = TRUE,
  ...
)

Arguments

mock_obj

Mock object that represents the distribution of measurements corresponding to mock samples.

list_gene

A character vector specifying the genes id to be retained for evaluation. If NULL (default) all genes are used for evaluation

list_tmb

TMB results from analysis.

dds

DESeq2 results from differential gene expression analysis.

coeff_threshold

A non-negative value which specifies a ln(fold change) threshold. The Threshold is used for the Wald test to determine whether the coefficient (β) is significant or not, depending on alt_hypothesis parameter. Default is 0.69, ln(FC = 2).

alt_hypothesis

Alternative hypothesis for the Wald test (default is "greaterAbs"). Possible choice: "greater"

  • β > coeff_threshold, "less"

  • β < −coeff_threshold, or two-tailed alternative: "greaterAbs"

  • |β| > coeff_threshold

alpha_risk

parameter that sets the threshold for alpha risk level while testing coefficient (β). Default: 0.05.

palette_color

Optional parameter that sets the color palette for plots.Default : c(DESeq2 = "#500472", HTRfit ="#79cbb8").

palette_shape

Optional parameter that sets the point shape for plots.Default : c(DESeq2 = 17, HTRfit = 19).

skip_eval_intercept

indicate whether to calculate precision-recall and ROC metrics for the intercept (default skip_eval_intercept = TRUE).

use_initial_random_params

A logical value indicating whether to use initial population values (TRUE) or simulated values (FALSE) for computing the actual sd values of random effects. Recommended usage : TRUE

...

Additional parameters to be passed to aesthetics get_pr_curve and get_roc_curve.

Value

A list containing the following components:

identity

A list containing model parameters and dispersion data.

precision_recall

A PR curve object generated from TMB and DESeq2 results.

roc

A ROC curve object generated from TMB and DESeq2 results.

counts

A counts plot generated from mock object.

performances

A summary of the performances obtained.

Examples

if (FALSE) {
report <- evaluation_report(list_tmb = l_res, dds = NULL, 
                           mock_obj = mock_data, 
                           coeff_threshold = 0.45, 
                           alt_hypothesis = "greaterAbs")
}