get_pvalue: The calculated \logchi^2 p-value for the fit of the elbow...

In ELBOW: ELBOW - Evaluating foLd change By the lOgit Way

Description

The following function determines a p value from the comparison of the slope of a regression line (i.e. logit). The function calculates a fold value from a set of initial conditions (replicate set 1) substracted from a set of final conditions (replicate set 2). The glm function is used to assess the fit of the dataset to the model, assuming family = binomial (link = "probit"), maxit = 1000. Determining the \logχ^2 p-value for the model, null equals no significance for the set of significant probes to the model, in which case the Elbow method cannot be used to assess the dataset. If significant probes are important to the model, then the \logχ^2 p-value will be below 0.05, in which case the Elbow method can be used to assess the dataset.

Usage

  get_pvalue(my_data, upper_limit, lower_limit)

Arguments

my_data	A table (data.frame) to analyze the elbow variance of. The columns in the table should be as follows: probes — one column containing the names of the probes fold — the fold values for the table
upper_limit	the upper limit/cut-off for the elbow.
lower_limit	the lower limit/cut-off for the elbow.

Value

The calculated \logχ^2 p-value for the elbow curve.

Examples

# read in the EcoliMutMA sample data from the package
		data(EcoliMutMA, package="ELBOW")
		csv_data <- EcoliMutMA
		# - OR - Read in a CSV file (uncomment - remove the #'s
		#        - from the line below and replace 'filename' with
		#        the CSV file's filename)
		# csv_data <- read.csv(filename)

		# set the number of initial and final condition replicates both to three
		init_count  <- 3
		final_count <- 3

		# Parse the probes, intial conditions and final conditions
		# out of the CSV file.  Please see: extract_working_sets
		# for more information.
		#
		# init_count should be the number of columns associated with
		#       the initial conditions of the experiment.
		# final_count should be the number of columns associated with
		#       the final conditions of the experiment.
		working_sets <- extract_working_sets(csv_data, init_count, final_count)

		probes <- working_sets[[1]]
		initial_conditions <- working_sets[[2]]
		final_conditions <- working_sets[[3]]

		# Uncomment to output the plot to a PNG file (optional)
		# png(file="output_plot.png")

		my_data <- replicates_to_fold(probes, initial_conditions, final_conditions)

		# compute the elbow for the dataset
		limits <- do_elbow(data.frame(my_data$fold))

		plus_minus <- elbow_variance(probes, initial_conditions, final_conditions)

		max_upper_variance <- plus_minus$max_upper
		min_upper_variance <- plus_minus$min_upper
		max_lower_variance <- plus_minus$max_lower
		min_lower_variance <- plus_minus$min_lower

		# rounded number for nice appearance graph
		upper_limit <- round(limits[[1]],digits=2)

		# rounded number nice appearance for graph
		lower_limit <- round(limits[[2]],digits=2)

		p_limits <- null_variance(my_data, upper_limit, lower_limit, initial_conditions)

		prowmin <- p_limits[[1]]
		prowmax <- p_limits[[2]]
		prowmedian <- p_limits[[3]]

		pvalue1 <- get_pvalue(my_data, upper_limit, lower_limit)

ELBOW documentation built on Nov. 8, 2020, 8:14 p.m.