Skip to main content
Topic: GC-TOF data processing in MSDIAL and redundancies in features identification (Read 3465 times) previous topic - next topic

GC-TOF data processing in MSDIAL and redundancies in features identification

Hello,

We collected metabolomics data using an Agilent GC coupled to a LECO HRT+ TOFMS. We then exported the data as netCDF, converted to .ABF format, and processed in MS-DIAL 4.70. We used a FAME mix to set the Retention Index to aid in peak identification. We were able to obtain library-matched features, but after checking the alignment table, we found redundancies of features identified. That is, the redundant features have the same retention time, same quant mass and are consecutively repeated 3 to 4 times. Additionally, some features have slight differences in retention times (~0.01 minutes) and quant masses but identified as the same metabolite. Reviewing the XIC's for such features shows these are unlikely to be different molecules but poorly aligned. I've attached a snapshot of the alignment table to illustrate what we're seeing. This is the first time we are using MSDIAL to process a large dataset from this LECO HRT+ instrument (we have used it in the past with single quad GC-MS & extensively with LC-MS data without such issues). We suspect a setting in the identification and/or alignment windows to be the culprit of the redundancies, but would appreciate some guidance and suggestions as to how to solve this issue.

Many thanks in advance. We're happy to share our specific processing settings if that would be helpful.

Banani


Re: GC-TOF data processing in MSDIAL and redundancies in features identification

Reply #1
Dear Banani,

sounds like the alignment is too strikt with the accurate mass.

I dont have access to an accurate gc-ms, but I know that the alignment settings can be changed for accurate mass for lc-ms but there is no mass deviation setting for normal gc-ms.
Maybe its missing for accurate gc mass or hidden in the tab where you can click accurate mass (peak detection)?

Best,
Martin

Re: GC-TOF data processing in MSDIAL and redundancies in features identification

Reply #2
Hi Martin,

Thanks for the suggestion, and yes we have the accurate mass box checked.


Banani

Re: GC-TOF data processing in MSDIAL and redundancies in features identification

Reply #3
Hi,
 I have the same problem with nominal GC-MS data. It seems tricky to have perfect alignment parameters. What I have done was to quickly turn those replicates into unknowns by Ctrl + D in the ion table.

Best,
Sukis

Re: GC-TOF data processing in MSDIAL and redundancies in features identification

Reply #4
Hi all,

I actually was addressing this problem for lcms data but not for gcms.
I will check the gcms project source code as soon as possible. Sorry for the inconvenience.

Let me ask one question. Do you use the accurate mass setting? Your quant mass data seems to be nominal mass.

Hiroshi

Re: GC-TOF data processing in MSDIAL and redundancies in features identification

Reply #5
Hello Dr. Hiroshi,

Thanks for your reply and yes, we used accurate mass setting.
Also, once I checked the box "only report the top hit" under identification tab, removes the redundancies of many features but not all. I was planning to filter the ion table in excel for the existing duplicates. 

Thanks,
Banani

Re: GC-TOF data processing in MSDIAL and redundancies in features identification

Reply #6
Banani,

Please share the Parameters.text file and I can take a look for HRGCMS settings for which I can suggest optimal settings....such as ToF or Orbitrap...

Thanks,
Biswa

Re: GC-TOF data processing in MSDIAL and redundancies in features identification

Reply #7
Hello Biswapriya,

I have shared the parameters setting files screenshot here if that helps. I can save the parameter file only as Parameter.med2 but can not upload that type of file here. It would be great to get suggestions from you.

Thanks,
Banani

 

Re: GC-TOF data processing in MSDIAL and redundancies in features identification

Reply #8
Hi, I have no experience with Fiehn RI, but is the tolerant RI 3000 for identification and alignment too big?

Best,
Sukis

Re: GC-TOF data processing in MSDIAL and redundancies in features identification

Reply #9
Hi all,    
    I am guessing that the problem could lie in the deconvolution step. As shown below in the Figure, a tiny peak (45.1 m/z) near the main peak (diphenyl ether) has a bad deconvoluted MS spectrum as it is heavily affected by diphenyl ether. Hence, it has almost the same spectrum as diphenyl ether and of course, it is identified as diphenyl ether with a high matching score. In the end, both the major peak (real diphenyl ether) and the tiny peak (45.1 m/z) will be used for alignment and cause a redundant alignment table.
   As a comparison, the tiny peak (45.1 m/z) seems to have a better-deconvoluted spectrum by AMDIS (not strongly affected by diphenyl ether). For this reason, I am guessing a way to go is to improve the deconvoluted spectrum of the tiny peak. I have tried to apply different sigma values for the deconvolution step, but unfortunately, it did not help.
   I am thinking if it is possible to calculate the quality of every deconvoluted spectrum, then we can apply a threshold to filter out all badly deconvoluted spectra for annotation as they are meaningless.

Best regards,
Qizhi Su

Re: GC-TOF data processing in MSDIAL and redundancies in features identification

Reply #10
Hi Qizhi and all,

thank you so much for this long time discussion. I am very happy to look at your example data, and if possible, I would like to improve the function. Thanks,

Hiroshi