error upon peak picking optimisation

[sneumann]

Hi,

I am getting an "object 'x1' not found" in a traceback(), see below. For debugging, I have restricted the parameters to be optimised to only ppm. This happens regardless whether or not I set nSlaves. What do I have to check to debug this further ?

Yours, Steffen

peakpickingParameters <- getDefaultXcmsSetStartingParams('centWave')

if (DEBUG) {
   ## reduce what gets optimised
   peakpickingParameters$min_peakwidth <- 20
   peakpickingParameters$max_peakwidth <- 20
   peakpickingParameters$mzdiff <- 0
   peakpickingParameters$ppm <- c(30,60)
}

iporesult <- optimizeXcmsSet(files=msfile,
                             params=peakpickingParameters,
                             subdir="/tmp")

Error in lapply(X = X, FUN = FUN, ...) : object 'x1' not found
> 
> traceback()
16: lapply(X = X, FUN = FUN, ...)
15: sapply(list(...), I)
14: FO(x1)
13: eval(expr, envir, enclos)
12: eval(predvars, data, env)
11: model.frame.default(formula = resp ~ FO(x1) + +PQ(x1), data = design, 
        drop.unused.levels = TRUE)
10: stats::model.frame(formula = resp ~ FO(x1) + +PQ(x1), data = design, 
        drop.unused.levels = TRUE)
9: eval(expr, envir, enclos)
8: eval(mf, parent.frame())
7: lm(formula = resp ~ FO(x1) + +PQ(x1), data = design)
6: eval(expr, envir, enclos)
5: eval(CALL, parent.frame())
4: rsm(formula, data = design)
3: createModel(xcms_result$design, params$to_optimize, resp)
2: xcmsSetStatistic(xcms_result, subdir, iterator)
1: optimizeXcmsSet(files = msfile, params = peakpickingParameters, 
       subdir = paste(outdir, basename(msfile), sep = "/"))

[sneumann]

Gunnar suggested that the optimisation requires at least to parameters to be optimised. So to fail gracefulle, we need to check " length(which(sapply(peakpickingParameters, length)>1))>1" at the beginning of optimizeXcmsSet() (and likely also grouping etc)

[glibiseller]

I'll soon add a new version which simply tests different levels and uses the best level as center for the next DoE.

[glibiseller]

The branch supportSingleParameterOptimization contains support for single parameter optimization. Since only one parameter is optimized within each DoE nine different levels are tested (instead of three levels CCD tests).

e.g: level <- c(10,20) seq(min(level), max(level), diff(level)/8)

This simple approach leads to redundant level testing in consecutive DoEs which should be avoided in a final version.

[glibiseller]

merged into master branch