Skip to main content
Topic: Mass Accuracy Seetings (Read 4368 times) previous topic - next topic

Mass Accuracy Seetings

Hello MS-DIAL Developers,

In the MS-DIAL Results (output files), I see that the 'Quant mass' column does not have 5 decimal places (mostly 2 decimals and rarely 2, random?) covered for the  HRGC-EI-MS data, though the 'EI-spectrum column' covers all 5 decimal places. And I checked its not hit/ library specific and very random.

Please see the attached snapshot.

Is there a setting while exporting results to specify for consistency?

Thanks,
Biswa

Re: Mass Accuracy Seetings

Reply #1
Thanks Biswa,

I will fixed it by the next update, and the new version of msdial (version 4.02) will be released in next week.
Thanks!

Hiroshi

Re: Mass Accuracy Seetings

Reply #2
Hi Biswa,

I would like to ask you several questions about the issue of quant mass decimals.
(1) Which msp library did you use?
(2) If you used something from my website (http://prime.psc.riken.jp/Metabolomics_Software/MS-DIAL/), note that this contains "QuantMass" field in msp.
(3) Did you use"Use quant masses defined in MSP format file" option which is available at 'identification' tab of parameter setting in MS-DIAL?

Most of quant masses that I created in the MSP files are recorded by nominal or 2 decimal values, and the decimal values will be reflected in your result for quant masses as well. Please confirm it first because I could not reproduce your result in my data set from GC HR-MS.

Thanks,

Hiroshi 

Re: Mass Accuracy Seetings

Reply #3
Hi Hiroshi,

Thanks for the responses!
(1)   Your public libraries (all pooled from RIKEN, MoNA: all EIs) + my in house HR Orbitrap Libraries, of course, all pooled.
(2)   Yes, all from your RIKEN website, plus in house libraries. I see that the exact masses (see attached picture of 2 examples from EI-B and Fiehn-EI spectra) that in deed the masses are nominal mass, and the Exact Mass is several decimal places.
(3)   Yes, I used the “use quant masses defined in the MSP format file”. So question now is: If I uncheck this (= cross/NO), the Quant Masses column will be empty or will be those form the Query (and as I use HRMS spectra, this would be fine!)?

No wonder, I got a mix of nominal and accurate mass lists in the results.

I understand the nominal masses/ 2 decimal values for pre-existing libraries of the past or those available as open source.
I guess I now understand the issue, and unchecking the ‘quant mass defined in the MSP format file” is my solution?

Thanks a lot,
Biswa

Re: Mass Accuracy Seetings

Reply #4
Hi Biswa,

About: (2)   Yes, all from your RIKEN website, plus in house libraries. I see that the exact masses (see attached picture of 2 examples from EI-B and Fiehn-EI spectra) that in deed the masses are nominal mass, and the Exact Mass is several decimal places.

See attached picture. The MSP files on my website will have "QUANTMASS:" field that is used for quantifying the metabolite by the quant mass value +/- user-defined mass accuracy if you ticked the option of "Use quant masses defined in MSP format file".
I hope, your problem is resolved if you do not use this option actually. Or, you may replace the quant mass field with an accurate mass by your definition although it's not easy actually...

Hiroshi

Re: Mass Accuracy Seetings

Reply #5
Hi Hiroshi,
Thanks a lot for helping with these smaller yet significant issues!: )
I understand, adding “Exact masses” one by one or by calculation, either ways are very unhelpful!

I have 2 updates for you based on the above:

(1)   Firstly, as you see the attached snapshot, keeping the “Quant masses  Defined in msp format file OFF”  did not help (and actually got worse, and lost all the previously “displaying” HR exact masses too!). And as before, the “EI spectrum” columns are all AMs.

(2)   It has a decent (lowering # of hits!) Impact on the “performance” of identification ( if you kindly remember, I am blasting HRMS spectra against a mix of HRMS spectra plus unit resolution spectra!)  as follows:

ID vs Alignment Priority   Total                    Annotated         Unknown    Noise
Use Quant Masses Defined in MSP Format File ON     437   235   184   125
Use Quant Masses Defined in MSP Format File OFF   396   185   183   101

      
(Note: Do not add the numbers above as I have different naming conventions etc., but the performance in terms of matches went low with OFF!)

As a NOTE to future/ current  MS-DIAL users, I can attest easily, that keeping any of the options OFF is resulting in less # of hits in terms of annotated IDs and match quality, i.e.,

Use RT for scoring ON
Use RT Info for Filtering ON
Only report the top hit ON
Use Quant Masses Defined in MSP Format File ON
Identification After Alignment ON

See what makes sense and please advice for (1), (2) or we have to switch to the format where we OD have Quant Mass clearly specified: which is a big task for in house spectrum (time intensive!) and other open source sources! : )

Thanks again,
Biswa

Re: Mass Accuracy Seetings

Reply #6
Hi Biswa,

thanks, I actually checked my source codes, and I could resolve this issue.
1. please download the latest version at http://prime.psc.riken.jp/Metabolomics_Software/MS-DIAL/index2.html.
2. The high resolution's quant mass should always be displayed if you do not use "Use quant masses defined in MSP format file".
3. If you use the option of "Use quant masses defined in MSP format file", the quant mass becomes nominal if the record has Quant mass field by nominal mass.

Also, as you find, the option "Use quant masses defined in MSP format file" will affect the alignment result.
MS-DIAL provides the two dimensional alignment spots (RT vs Quant mass) as the result of peak alignment, but in the data processing, the alignment spots which have very similar RT and Quant mass values will be finalized so that one representative alignment spot is stored. Namely, if there were three spots having such similar RT and Quant mass values, two of them will be erased in the alignment process. Therefore, if your MSP file has many same quant mass values, the possibility to exclude alignment spots becomes higher than the case that you have set different quant mass value for each metabolite.

Thanks,

Hiroshi

Re: Mass Accuracy Seetings

Reply #7
Hi Hiroshi,

Thanks a lot for resolving the issues on a fast track mode and promptly so!

"2. The high resolution's quant mass should always be displayed if you do not use "Use quant masses defined in MSP format file". :  This should be so useful in 'fragment matching/ prediction'  etc. which was lost in the final  results for whatever reasons! : )

I will give a try on the 'quant mass ON' and see if it's AM or not in future runs esp. when we have unique Quant Mass for each spectrum from the in house HR EI-MS library .

This is good explanation of how the function  works and so useful in understanding!

Thanks again,
Biswa

 

Re: Mass Accuracy Seetings

Reply #8
Hi Hiroshi,

Thanks again!

As you can see in the screenshot (attached), that indeed that mass accuracy issue has been resolved for the Quant Mass for all the runs/ analysis irrespective of whether Quant Mass is specified during the analysis or not- in the new MS-DIAL version 4.1! : )

This is great!!!!

Thanks again,
Biswa