refitwoOutlier: Using subset of rows in a matrix to refit NLS model
In mengchen18/proturn: System-wise protein turnover curve fitting

Description Usage Arguments Value Examples

Given an output of fitNLSModels, using a subset of rows in the original matrix to fit a new NLS model. This functional is used by shinyapp, not for users.

refitwoOutlier(x, include, t, tcc, A = NULL, B = NULL, k = NULL,
  par.init = list(A = 1, B = 0.2, k = 0.04), par.lower = c(A = 0, B = 0, k =
  0), par.upper = c(A = 1, B = 1, k = 10), fitIndividual = TRUE,
  message = TRUE)

`x`	a model (i.e. an element of `list`) returned by `fitNLSModels`
`include`	which rows should be included
`t`	The time point (hours)
`tcc`	The doubling time of cells. By default this value is Inf, which means the cells are in steady state
`A`	optinal argument for fixed A, if this argument is given, "A" won't be optimized
`B`	optional argument for fixed B, if this argument is given, "B" won't be optimized
`k`	the synthesis/degradation constant
`par.init`	The initial values of parameters to be optimized, it should be list of three elements names as "A", "B" and "k".
`par.lower`	The lower boundary of parameters to be optimized, it should be a numeric values with length 3 and named as "A", "B" and "k".
`par.upper`	The upper boundary of parameters to be optimized, it should be a numeric values with length 3 and named as "A", "B" and "k".
`fitIndividual`	A logical value. When multiple lines are combined to fit a single model (see f), whether each individual lines should also be fitted. Only used when x is an object of class `matrix`
`message`	A logical value to indicated if any message should be printed

a vector of optimized parameters, including A, B, kd, confidence intervals (2.5 mean square error and r-square values. In addition, if individual rows are fitted, the object also contains an attribute stores parameters fitted on each individual row.

tp <- c(0, 1, 2, 4, 8, 16, 32, 64)
A <- runif(5, min = 0.75, max = 0.95)
B <- runif(5, min = 0.1, max = 0.15)
kd <- runif(5, min = 0.05, max = 0.1)
ds <- mapply(function(A, B, kd) {
  degCurve(A=A, B=B, kd=kd, tcc=Inf, t = tp) + rnorm(length(tp), sd = 0.03)
}, A, B, kd)
ds <- t(ds)
ds <- rbind(ds, runif(8, 0, 1))
rf <- c("p1", "p2", "p2", "p3", "p3", "p3")
ss <- fitNLSModels(x = ds, f = rf, t = tp, tcc = Inf, type = "deg",
                   par.init = list(A=0.8, B=0.2, kd=0.04),
                   par.lower=c(A=0, B=0, kd=0),
                   par.upper=c(A=1, B=1, kd=10))

plotCurve.comb(x = ss$list$p3, t = tp, tcc = Inf, curve = "deg")
#' only use the first two rows
res <- refitwoOutlier(x = ss$list$p3, include = 1:2, t = tp, tcc = Inf)
plotCurve.comb(x = res, t = tp, tcc = Inf, curve = "deg")


# synthesis curve
ds2 <- mapply(function(A, B, kd) {
  synCurve(A=A, B=B, kd=kd, tcc=Inf, t = tp) + rnorm(length(tp), sd = 0.03)
}, A, B, kd)
ds2 <- t(ds2)
ds2 <- rbind(ds2, runif(8, 0, 1))
ss2 <- fitNLSModels(x = ds2, f = rf, t = tp, tcc = Inf, type = "syn",
                    par.init = list(A=0.8, B=0.2, ks=0.04),
                    par.lower=c(A=0, B=0, ks=0),
                    par.upper=c(A=1, B=1, ks=10))

plotCurve.comb(x = ss2$list$p3, t = tp, tcc = Inf, curve = "syn")
#'#'
res <- refitwoOutlier(x = ss2$list$p3, include = 1:2, t = tp, tcc = Inf)
plotCurve.comb(x = res, t = tp, tcc = Inf, curve = "syn")