Skip to main content
Topic: MS-Finder search parameters (Read 3831 times) previous topic - next topic

MS-Finder search parameters

Hello everyone,

I just posted my first topic where I am asking for suggestions about Software for untargeted metabolomics (http://www.metabolomics-forum.com/index.php?topic=1572.new#new). As described in that topic, I already tried several common used softwares like MS-Dial and MS-Finder. The problem for me with MS-Finder is that I do not really trust those results... The main reason is that often the scores of the structure finder results are very low even though I adjusted the minimal score value to be 6 in the parameter settings.

Are any of you using MS-Finder for identification purpose and could anyone help me with this problem?

Leon

Re: MS-Finder search parameters

Reply #1
Hi,

ms-finder and the other related tools like sirius and gnps are the program to facilitate annotation for unknown spectra. And therefore, the annotation result of such structure elucidation programs should contain many false annotations, and users have to deal with the data by their own ways (biology and chemistry background knowledge etc).
The best practice to annotate metabolite is to use the experimental spectral libraries from massbank, gnps, and mona. And you can use such a library kit to be used in MS-DIAL program.
http://prime.psc.riken.jp/compms/msdial/main.html#MSP

Thanks,

Hiroshi

Re: MS-Finder search parameters

Reply #2
Thank you for your reply Hiroshi.

I know that the risk of getting false annotations is very high but that is why i wanted MS-Finder to only respond with compounds that have at least a score of 6 which i set as my limit. The problem is that MS-Finder does not filter these results and therefore results with a score lower than 6 are getting displayed.
Maybe i should give sirius a try again and also try out the HMDB from mona in MS-Dial instead of just only in Skyline...
I am fearing that other libraries may not contain all metabolites that are included in the sample or that other compounds (that just can not be in the sample or do not make sense) in the libraries falsely match some signals.
Is it also possible to do the quantification with MS-Dial?

Best,

Leon

 

Re: MS-Finder search parameters

Reply #3
Quick update

I tried to use the HMDB library (.msp, provided from mona) for identification with MS-Dial. No matter how low I am going with the identification parameters, MS-Dial displays 0 matched features.
I already ran into a similar problem before where MS-Dial would not match any features when using the provided libraries (http://prime.psc.riken.jp/compms/msdial/main.html#MSP). So I switched from a German to an English OS which then worked fine.
With the HMDB however neither German or English OS provide any matched features within MS-Dial.

So if anyone knows how to use the HMDB in MS-Dial for identification I would be glad to hear about it :)



Re: MS-Finder search parameters

Reply #4
Do you use HMDB library for GC-MS data? Or, do you try to use HMDB library for LC-MS?
In HMDB library, there are few information for LC-MS/MS, and there is no precursor m/z field which is required for annotation in LC-MS/MS project in MS-DIAL.
I simply downloaded HMDB msp library from mona website, and opened it by notepad++. Mostly, the contents are for GC-MS as attached.
Thanks,

Hiroshi

Re: MS-Finder search parameters

Reply #5
Hi Hirsoshi,

I was using the library for LC-MS and also noticed those few entries for LC-MS..
However just a few days ago I tried to do Identification for some spiked samples with the provided GNPS-Library. Out of the samples which were measured with a DIA method I received about 100 identified metabolites from over thousand(s) of features. Even when I "lowered" the identification parameters the number of identified metabolites did not change. Also, if I remember correctly, not all of the spiked metabolites were identified. Maybe you do already have an idea of what could be the reason for this with those information. Nevertheless I can post my parameters as soon as possible to give you more details.

Best,

Leon

Re: MS-Finder search parameters

Reply #6
Hi Hirsoshi,

I was using the library for LC-MS and also noticed those few entries for LC-MS..
However just a few days ago I tried to do Identification for some spiked samples with the provided GNPS-Library. Out of the samples which were measured with a DIA method I received about 100 identified metabolites from over thousand(s) of features. Even when I "lowered" the identification parameters the number of identified metabolites did not change. Also, if I remember correctly, not all of the spiked metabolites were identified. Maybe you do already have an idea of what could be the reason for this with those information. Nevertheless I can post my parameters as soon as possible to give you more details.

Best,

Leon

For the identification I used a tolerance of 0.1 Da each and a 50% score cut off. This should definitely give me more than my 92 identified features...
What also came in mind when I reviewed the data today was that even though the DIA data should contain MS2 spectra for 75 Da to 325 Da (window width is evenly distributed: 75-125, 125-175,...) MS-DIAL only displayed a range from 75 Da to about 140 Da. I do not know why that is the case.. My experiment file is just as the one that is provided and I also did not set any m/z limitation in the settings.

It would be such a relief to know what the reasons for my failure with MS-DIAL are :(

Re: MS-Finder search parameters

Reply #7
>>What also came in mind when I reviewed the data today was that even though the DIA data should contain MS2 spectra for 75 Da >>to 325 Da (window width is evenly distributed: 75-125, 125-175,...) MS-DIAL only displayed a range from 75 Da to about 140 Da.

Maybe, this is due to either (A) experimental file problem or (B) the converted file problem.
What is your instrument used? SCIEX? SWATH-MS? or the others?
If sciex, what is your raw data format type (wiff or wiff2)?
Please let me know your experimental/analysis detail, and if possible, please send some example data to me.
Thanks,

Hiroshi

Re: MS-Finder search parameters

Reply #8
Hi Leon

about this>>I know that the risk of getting false annotations is very high but that is why i wanted MS-Finder to only respond with compounds that have at least a score of 6 which i set as my limit. The problem is that MS-Finder does not filter these results and therefore results with a score lower than 6 are getting displayed.

There was a bug where the score cut off of msfinder structure elucidation did not work well.
I fixed it, and the new version will be uploaded soon.

Hiroshi


Re: MS-Finder search parameters

Reply #9
>>What also came in mind when I reviewed the data today was that even though the DIA data should contain MS2 spectra for 75 Da >>to 325 Da (window width is evenly distributed: 75-125, 125-175,...) MS-DIAL only displayed a range from 75 Da to about 140 Da.

Maybe, this is due to either (A) experimental file problem or (B) the converted file problem.
What is your instrument used? SCIEX? SWATH-MS? or the others?
If sciex, what is your raw data format type (wiff or wiff2)?
Please let me know your experimental/analysis detail, and if possible, please send some example data to me.
Thanks,

Hiroshi

Hello Hiroshi,

I have already send you some example data about 1-2 months ago. Could you tell me if you have received these or if you are even working on this problem?

Thank you,

Leon


Re: MS-Finder search parameters

Reply #10
Hi Leon,

today, I had a time to see your data.
I do not know the reason, but your thermo-ms file contains the ms1/ms2 cycle as the following order:
CycleID   Type   m/z begin   m/z end
0   SWATH   75   108
1   SWATH   108   141
2   SWATH   141   174
3   SWATH   174   207
4   SWATH   207   240
5   SWATH   240   273
6   SCAN   75   273

I attached the experimental file that I modified. If you use this, you will be able to process your data correctly.
You can confirm this order actually by using SeeMS from proteowizard community.

Thanks,

Hiroshi