msmsRead fails

[stanstrup]

Hi,

I am trying to process some old MS/MS runs but running into an error I cant seem to pin down. It seems to come down to it looking for a RT at some point, not finding it, returning an empty object and failing.

I have attached the raw file and the script to reproduce the error.

> library(RMassBank)
> 
> file <- "2012-07-11-033.mzData"
> 
> loadRmbSettings("RMB_options.ini")
> w <- newMsmsWorkspace(file)
> loadList("Compoundlist.csv")
Loaded compoundlist successfully
> 
> 
> w <- msmsRead(w, mode="pH", files=file, cpdids = 6, readMethod="xcms") # mzR errors too
  |                                                                                                                                                     |                                                                                                                                             |   0%Error in if (.listEnvEnv$listEnv$compoundList[match(cpdID, .listEnvEnv$listEnv$compoundList$ID),  : 
  missing value where TRUE/FALSE needed

R version 3.3.1 (2016-06-21)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows >= 8 x64 (build 9200)

locale:
[1] LC_COLLATE=Danish_Denmark.1252  LC_CTYPE=Danish_Denmark.1252    LC_MONETARY=Danish_Denmark.1252 LC_NUMERIC=C                   
[5] LC_TIME=Danish_Denmark.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] RMassBank_2.2.0     Rcpp_0.12.5         RevoUtilsMath_8.0.3

loaded via a namespace (and not attached):
 [1] lattice_0.20-33        png_0.1-7              digest_0.6.9           foreach_1.4.3          chron_2.3-47           plyr_1.8.4            
 [7] acepack_1.3-3.3        mzID_1.12.0            stats4_3.3.1           ggplot2_2.1.0          BiocInstaller_1.24.0   itertools_0.1-3       
[13] zlibbioc_1.20.0        data.table_1.9.6       S4Vectors_0.12.0       rpart_4.1-10           Matrix_1.2-6           preprocessCore_1.36.0 
[19] mzR_2.8.0              xcms_1.50.0            splines_3.3.1          RevoUtils_10.0.1       BiocParallel_1.8.1     stringr_1.0.0         
[25] foreign_0.8-66         ProtGenerics_1.6.0     igraph_1.0.1           RCurl_1.95-4.8         munsell_0.4.3          fingerprint_3.5.2     
[31] BiocGenerics_0.20.0    multtest_2.30.0        pcaMethods_1.66.0      nnet_7.3-12            gridExtra_2.2.1        RANN_2.5              
[37] Hmisc_3.17-4           IRanges_2.8.0          codetools_0.2-14       XML_3.98-1.4           MASS_7.3-45            bitops_1.0-6          
[43] grid_3.3.1             MassSpecWavelet_1.40.0 RBGL_1.50.0            CAMERA_1.30.0          gtable_0.2.0           affy_1.52.0           
[49] magrittr_1.5           rcdk_3.3.7             scales_0.4.0           graph_1.52.0           stringi_1.1.1          impute_1.48.0         
[55] reshape2_1.4.1         affyio_1.44.0          doParallel_1.0.10      limma_3.30.2           latticeExtra_0.6-28    Formula_1.2-1         
[61] rjson_0.2.15           RColorBrewer_1.1-2     iterators_1.0.8        tools_3.3.1            Biobase_2.34.0         MSnbase_2.0.0         
[67] rcdklibs_1.5.13        parallel_3.3.1         survival_2.39-5        yaml_2.1.13            colorspace_1.2-6       cluster_2.0.4         
[73] vsn_3.42.3             MALDIquant_1.15        rJava_0.9-8           

error_hunt.zip

[meowcat]

Hi,

I just checked. While I don't know yet what the error is, two things

• you use mzData. I have last used that format years ago and never tested RMassBank with it anymore - we use mzML and mzXML. I don't know if the mzData parser for mzR is bad or if the data is not in the file, but when you open the file by hand with mzR and look at the header, there are no info about collision energy etc.

library(mzR)
d <- openMSfile(file)
h <- header(d)
h[110:130,]

• Connected to this, there are no MS1 scans in the file. In this case reading with the mzR option will not work (since it needs the MS1 scans to group MS2 scans together, and to verify the presence of the precursor).
• On the other hand, I think it could still work with the XCMS option (I have to look up how precisely that works) but there are no collision energies in the file, and if I remember right XCMS needs collision energies to split the data correctly.

So my first suggestion would be to see if converting to mzML or mzXML helps at all (i.e. gives you collision energies in the file). It could work with the XCMS option then.

@ermueller: Does the xcms option still require an exact retention time? Because one could also build a pseudo-EIC from the MS2 TIC (per collision energy) and pick the RT from there, if the user is confident that this will be enough to pick the standard.

[stanstrup]

Hi,

thanks for looking into it. I did notice the collision energy thing but when I tried to bug-hunt it seemed irrelevant since the collision energy was just set to 0. In this case there was only 1 energy used so it didn't seem to make a difference. I now reconverted to mzML and it has the collision energy but the result is the same. I have attached the new file.

2012-07-11-033_lockmass_rem.zip

[meowcat]

Hi,

sorry for the low responsiveness; I tend to be busy with other stuff currently...

Multiple things here:

• mzR will still not work because there are no precursor scans... this is either "future features" or "will not fix", to be decided. A fix would effectively be a new or additional logic of how to look for the spectra.
• xcms workflow had a bug for the "standard" case where the number of compound IDs matches the number of files (instead of 1 file - multiple compounds). This led to the error you observed. Fixed in 1d86138
  • xcms workflow, as designed currently, requires the RT in the compound list. There could be ways around this (to be discussed with @ermueller) but to some degree is also "future features".
  • If I add the RT, a spectrum is retrieved. However it seems to be the wrong one (the correct RT seems to be 1.44 whereas a near-empty spectrum at 1.27 is returned.) Compoundlist_RT.zip To be followed up.