haplo.stats (version 1.7.9)

score.sim.control: Create the list of control parameters for simulations in haplo.score

Description

In the call to haplo.score, the sim.control parameter is a list of parameters that control the simulations. This list is created by this function, score.sim.control, making it easy to change the default values.

Usage

score.sim.control(p.threshold=0.25, min.sim=1000, max.sim=20000.,verbose=FALSE)

Arguments

p.threshold

A paremeter used to determine p-value precision from Besag and Clifford (1991). For a p-value calculated after min.sim simulations, continue doing simulations until the p-value's sample standard error is less than p.threshold * p-value. The dafault value for p.threshold = 1/4 corresponds approximately to having a two-sided 95% confidence interval for the p-value with a width as wide as the p-value itself. Therefore, simulations are more precise for smaller p-values. Additionally, since simulations are stopped as soon as this criteria is met, p-values may be biased high.

min.sim

The minimum number of simulations to run. To run exactly min.sim simulations, set max.sim = min.sim. Also, if run-time is an issue, a lower minimum (e.g. 500) may be useful, especially when doing simulations in haplo.score.slide.

max.sim

The upper limit of simulations allowed. When the number of simulations reaches max.sim, p-values are approximated based on simulation results at that time.

verbose

Logical, if (T)rue, print updates from every simulation to the screen. If (F)alse, do not print these details.

Value

A list of the control parameters:

p.threshold

As described above

min.sim

As described above.

max.sim

As described above

verbose

As described above

Details

In simulations for haplo.score, employ the simulation p-value precision criteria of Besag and Clifford (1991). The criteria ensures both the global and the maximum score statistic simulated p-values be precise for small p-values. First, perform min.sim simulations to guarantee sufficient precision for the score statistics on individual haplotypes. Then continue simulations as needed until simulated p-values for both the global and max score statistics meet precision requirements set by p.threshold.

References

Besag, J and Clifford, P. "Sequential Monte Carlo p-values." Biometrika. 78, no. 2 (1991): 301-304.

See Also

haplo.score

Examples

# NOT RUN {
# it would be used in haplo.score as appears below
#
# score.sim.500 <- haplo.score(y, geno, trait.type="gaussian", simulate=T, 
#                sim.control=score.sim.control(min.sim=500, max.sim=2000)
# }

Run the code above in your browser using DataCamp Workspace