Error in extractChromMetrics

[pablovgd]

Dear William,

I was messing around with squallms and stumbled on this error which I'm not able to fix easily. I've used squallms this way before without errors. Any idea what might be going on? Thanks in advance!

Cheers, Pablo

Error in `summarise()`:
ℹ In argument: `beta_vals = list(qscoreCalculator(rt, int))`.
ℹ In group 4128: `feature = "FT0052"` and `filename = "230626s380_QC1_29.mzXML"`.
Caused by error in `cor()`:
! incompatible dimensions
Backtrace:
  1. squallms::extractChromMetrics(feat_peak_info_subset, ms1_data = msdata$MS1)
 17. squallms:::qscoreCalculator(rt, int)
 18. stats::cor(perf_peak, int)
Error in summarise(., peak_mz = weighted.mean(mz, int), peak_rt = median(rt),  : 

ℹ In group 4128: `feature = "FT0052"` and `filename = "230626s380_QC1_29.mzXML"`.
Caused by error in `cor()`:
! incompatible dimensions

I followed the steps in the vignette .rmd:

feat_peak_info <- makeXcmsObjFlat(result) %>%
    select(feature, starts_with("mz"), starts_with("rt"), filename, filepath)
feat_peak_info %>%
    head() %>%
    mutate(filepath = paste0(substr(filepath, 1, 13), "~")) %>%
    knitr::kable(caption = "Output from makeXcmsObjFlat.")
feat_peak_info_subset <- feat_peak_info %>%
    filter(grepl("QC", filename))
msdata <- grabMSdata(unique(feat_peak_info_subset$filepath), verbosity = 0)
shape_metrics <- extractChromMetrics(feat_peak_info_subset, ms1_data = msdata$MS1)

[wkumler]

Phew, good question. I'm not sure what would actually be causing this but I know it's because somehow the retention time and intensity vectors are different lengths (the same error shows up when you do cor(1:2, 1:3). Some debugging info might be necessary and if you're willing to test some things then I might be able to push a bug fix or help you get around it.

• Is it just that one feature? I.e. if you filter out FT0052 from the feat_peak_info_subset does it error on a different one?
• Does it still show up if you don't filter for just the QC files?
• If you activate debugging with options(error=recover) and navigate to the last frame (stats::cor(perf_peak, int)) then print x and y, what are they? (If you haven't used error=recover before you can get out of the Browse[1]> prompt by typing Q and hitting enter, then disabling it by setting the default again options(error=NULL))
• You might be able to get around it by rerunning the XCMS peakpicking step with verboseBetaColumns = TRUE and using the betas calculated there instead (you may have already done this so all you'd need to do is not drop those columns during the select step). Does XCMS throw an error if you do that?