invert.pmultinom {pmultinom}

Calculate the sample size such that the probability of a result is a given amount.

Description

Calculate the sample size such that the probability of a result is a given amount.

Usage

invert.pmultinom(lower = -Inf, upper = Inf, probs, target.prob, method)

Arguments

lower

Vector of lower bounds. Lower bounds are excluded

upper

Vector of upper bounds. Upper bounds are included

probs

Cell probabilities

target.prob

The probability of the event, at the output sample size.

method

Method used for computation. Only method currently implemented is "exact"

Details

If only lower is given, then the result is the smallest size such that pmultinom(lower=lower, size=size, probs=probs) >= target.prob. If only upper is given, then the result is the smallest size such that pmultinom(upper=upper, size=size, probs=probs) <= target.prob. Behavior when both lower and upper are given is not yet implemented.

Value

The sample size parameter at which the the target probability of the given event is achieved.

References

Casasent, A. K., Schalck, A., Gao, R., Sei, E., Long, A., Pangburn, W., ... & Navin, N. E. (2018). Multiclonal Invasion in Breast Tumors Identified by Topographic Single Cell Sequencing. Cell. doi:10.1016/j.cell.2017.12.007

See Also

pmultinom

Examples

# How many cells must be sequenced to have a 95% chance of
# observing at least 2 from each subclone of a tumor? (Data
# from Casasent et al (2018); see vignette("pmultinom") for
# details of this example)

# Input: 
ncells <- 204
subclone.freqs <- c(43, 20, 82, 17, 5, 37)/ncells
target.number <- c(2, 2, 2, 2, 2, 0)
lower.bound <- target.number - 1
invert.pmultinom(lower=lower.bound, probs=subclone.freqs,
                 target.prob=.95, method="exact")
# Output:
# [1] 192