edgeR (version 3.14.0)

goodTuring: Good-Turing Frequency Estimation

Description

Non-parametric empirical Bayes estimates of the frequencies of observed (and unobserved) species.

Usage

goodTuring(x, conf=1.96)
goodTuringPlot(x)
goodTuringProportions(counts)

Arguments

x

numeric vector of non-negative integers, representing the observed frequency of each species.

conf

confidence factor, as a quantile of the standard normal distribution, used to decide for what values the log-linear relationship between frequencies and frequencies of frequencies is acceptable.

counts

matrix of counts

Value

goodTuring returns a list with components

count

observed frequencies, i.e., the unique positive values of x

n

frequencies of frequencies

n0

frequency of zero, i.e., number of zeros found in x

proportion

estimated proportion of each species given its count

P0

estimated combined proportion of all undetected species

goodTuringProportions returns a matrix of proportions of the same size as counts.

Details

Observed counts are assumed to be Poisson distributed. Using an non-parametric empirical Bayes strategy, the algorithm evaluates the posterior expectation of each species mean given its observed count. The posterior means are then converted to proportions. In the empirical Bayes step, the counts are smoothed by assuming a log-linear relationship between frequencies and frequencies of frequencies. The fundamentals of the algorithm are from Good (1953). Gale and Sampson (1995) proposed a simplied algorithm with a rule for switching between the observed and smoothed frequencies, and it is Gale and Sampson's simplified algorithm that is implemented here. The number of zero values in x are not used in the algorithm, but is returned by this function.

Sampson gives a C code version on his webpage at http://www.grsampson.net/RGoodTur.html which gives identical results to this function.

goodTuringPlot plots log-probability (i.e., log frequencies of frequencies) versus log-frequency.

goodTuringProportions runs goodTuring on each column of data, then uses the results to predict the proportion of each gene in each library.

References

Gale, WA, and Sampson, G (1995). Good-Turing frequency estimation without tears. Journal of Quantitative Linguistics 2, 217-237.

Examples

#  True means of observed species
lambda <- rnbinom(10000,mu=2,size=1/10)
lambda <- lambda[lambda>1]

#  Oberved frequencies
Ntrue <- length(lambda)
x <- rpois(Ntrue, lambda=lambda)
freq <- goodTuring(x)
goodTuringPlot(x)