dpcR (version 0.5)

test_counts: Test counts

Description

The test for comparing counts from two or more digital PCR experiments.

Usage

test_counts(input, model = "ratio", conf.level = 0.95)

Arguments

input

object of class '>adpcr or '>dpcr with "nm" type.

model

may have one of following values: binomial, poisson, prop, ratio. See Details.

conf.level

confidence level of the intervals and groups.

Value

an object of class '>count_test.

Details

test_counts incorporates two different approaches to models: GLM (General Linear Model) and multiple pair-wise tests. The GLM fits counts data from different digital PCR experiments using quasibinomial or quasipoisson family. Comparisons between single experiments utilize Tukey's contrast and multiple t-tests (as provided by function glht).

In case of pair-wise tests, (rateratio.test or prop.test) are used to compare all pairs of experiments. The p-values are adjusted using the Benjamini & Hochberg method (p.adjust). Furthermore, confidence intervals are simultaneous.

References

Bretz F, Hothorn T, Westfall P, Multiple comparisons using R. Boca Raton, Florida, USA: Chapman & Hall/CRC Press (2010).

See Also

Functions used by test_counts:

• glm,

• glht,

• cld,

• prop.test,

• rateratio.test

GUI presenting capabilities of the test: test_counts_gui.

Examples

# NOT RUN {
#be warned, the examples of test_counts are time-consuming
# }
# NOT RUN {
adpcr1 <- sim_adpcr(m = 10, n = 765, times = 1000, pos_sums = FALSE, n_panels = 3)
adpcr2 <- sim_adpcr(m = 60, n = 550, times = 1000, pos_sums = FALSE, n_panels = 3)
adpcr2 <- rename_dpcr(adpcr2, exper = "Experiment2")
adpcr3 <- sim_adpcr(m = 10, n = 600, times = 1000, pos_sums = FALSE, n_panels = 3)
adpcr3 <- rename_dpcr(adpcr3, exper = "Experiment3")

#compare experiments using binomial regression
two_groups_bin <- test_counts(bind_dpcr(adpcr1, adpcr2), model = "binomial")
summary(two_groups_bin)
plot(two_groups_bin)
#plot aggregated results
plot(two_groups_bin, aggregate = TRUE)
#get coefficients
coef(two_groups_bin)

#this time use Poisson regression
two_groups_pois <- test_counts(bind_dpcr(adpcr1, adpcr2), model = "poisson")
summary(two_groups_pois)
plot(two_groups_pois)

#see how test behaves when results aren't significantly different
one_group <- test_counts(bind_dpcr(adpcr1, adpcr3))
summary(one_group)
plot(one_group)
# }