Create data with function inside shinyServer to feed ggplot - r

Question

I have a Shiny app what calculates some power estimates for a type of genetic association study. The ui.R is pretty simple, and the server.R has a function that gives a data frame (I think I can't have this function as reactive because it has some parameters).
The link to the Gist is here. To run it:
library(shiny)
shiny:: runGist('5895082')
The app calculates correctly the estimates, but I have two questions regarding it:
Is it possible to have the output$powTable actually represent all the values contained within the range, in the first sliderInput(n.cases)?. It only seems to represent the two extreme values of the range... what I'm doing wrong?
There's an error when running the app:
Error: Reading objects from shinyoutput object not allowed.
How can I pass the data (reactivity?) from the function f() to feed the ggplot? After much trial and error, I am very lost. Where can be the error in my code? Many thaks in advance!
The original code of the function works well: (EDITED)
f <- function(ncases, p0, OR.cas.ctrl, Nh, sig.level) {
num.cases <- ncases
p0 <- p0
Nh <- Nh
OR.cas.ctrl <- OR.cas.ctrl
sig.level <- sig.level
# Parameters related to sig.level, from [Table 2] of Samuels et al.
# For 90% power and alpha = .05, Nscaled = 8.5
if (sig.level == 0.05){
A <- -28 # Parameter A for alpha=.05
x0 <- 2.6 # Parameter x0 for alpha=.05
d <- 2.4 # Parameter d for alpha=.05
}
if (sig.level == 0.01){
A <- -13 # Parameter A for alpha=.01
x0 <- 5 # Parameter x0 for alpha=.01
d <- 2.5 # Parameter d for alpha=.01
}
if (sig.level == 0.001){
A <- -7 # Parameter A for alpha=.001
x0 <- 7.4 # Parameter x0 for alpha=.001
d <- 2.8 # Parameter d for alpha=.001
}
out.pow <- NULL # initialize vector
for(ncases in ncases){
OR.ctrl.cas <- 1 / OR.cas.ctrl # 1. CALCULATE P1 FROM A PREDEFINED P0, AND A DESIRED OR
OR <- OR.ctrl.cas
bracket.pw <- p0 / (OR - OR*p0) # obtained after isolating p1 in OR equation [3].
p1 <- bracket.pw / (1 + bracket.pw)
Nh037 <- Nh^0.37 # 2. CALCULATE NSCALED
num.n <- num.cases*((p1-p0)^2)
den.n <- (p1*(1-p1) + p0*(1-p0))*Nh037
Nscaled <- num.n/den.n
num.power <- A - 100 # 3. CALCULATE POWER
den.power <- 1 + exp((Nscaled - x0)/d)
power <- 100 + (num.power/den.power) # The power I have to detect a given OR with my data, at a given alpha
}
OR <- OR.cas.ctrl
out.pow <- data.frame(num.cases, Nh, Nscaled, p0, OR, sig.level, power)
out.pow
}
mydata <- f(ncases=seq(50,1000, by=50), 0.4, 2.25, 11, 0.05)
mydata
library(ggplot2)
print(ggplot(data = mydata, aes(num.cases, power)) +
theme_bw() +
theme(text=element_text(family="Helvetica", size=12)) +
labs(title = "Ad-hoc power for haplogroup") +
scale_color_brewer(palette = "Dark2", guide = guide_legend(reverse=TRUE)) +
xlab("number of cases/controls") +
ylab("power") +
scale_x_log10() +
geom_line(alpha=0.8, size=0.2) +
geom_point(aes(shape = factor(OR)), colour="black"))

Answer 1

First of all, you have n.cases named inconsistently I think. It's n.cases sometimes, and ncases other times. Is that a mistake?
Anyway, output$mydata() is incorrect. It isn't an output. It should be just:
mydata <- reactive(f(input$n.cases,
input$p0,
input$OR.cas.ctrl,
input$Nh,
input$sig.level))
And then when executing it in output$powHap() it should be:
output$powHap <- renderPlot(
{
print(ggplot(data = mydata(), aes(ncases, power)) +
theme_bw() +
theme(text=element_text(family="Helvetica", size=12)) +
labs(title = "Ad-hoc power for haplogroup") +
scale_color_brewer(palette = "Dark2", guide = guide_legend(reverse=TRUE)) +
xlab("number of cases/controls") +
ylab("power") +
scale_x_log10() +
geom_line(alpha=0.8, size=0.2) +
geom_point(aes(shape = factor(OR)), colour="black"))
})
The important part there is that you need to do:
data = mydata()
rather than
data = output$mydata
Because output$mydata is a (reactive) function.
I would suggest reading the documentation on how reactives work. The whole thing should make a lot more sense afterwards. +1 for a very reproducible example by the way. This is how all questions should be posted.