Combination of consensus libraries and import in MRM format via spectraST

42 views
Skip to first unread message

Kristina Plate

unread,
Sep 20, 2019, 9:22:28 AM9/20/19
to spctools-discuss
Hello again,

as I described before (topic: What happens during consensus building?) I have a problem with some messed up spectra in my spectral libraries. 
I combined four spectral libraries and built consensus spectra. Afterwards I reimported the combined consensus spectral library in MRM format and converted it in an assay library. When I checked the SPTXT file I found some slightly changed spectra and new "ion types" m. Please see my example of an old, normal spectrum and the corresponding new one below:

Name: DIETIIQK/2 (old entry, original library)
LibID: 4160
MW: 960.5481
PrecursorMZ: 480.2740
Status: Normal
FullName: X.DIETIIQK.X/2 (CID-QTOF)
Comment: AvePrecursorMz=480.5611 BinaryFileOffset=23385298 FracUnassigned=0.09,1/5;0.23,11/20;0.06,7/31 MassDiff=-0.0014 Mods=0 NAA=8 NISTProtein=Id_02 NMC=0 NTT=1 Nreps=1/1 OrigMaxIntensity=11 Parent=480.274 Pep=Tryptic PrecursorIntensity=0 Prob=1.0000 Protein=1/Id_02 RawSpectrum=file_02_04.69683.69683 RetentionTime=2745.2,2745.2,2745.2 Sample=1/sample_02,1,1 Se=1^C1:pb=1.0000/0,fv=2.7040/0 Spec=CREATE TotalIonCurrent=8.9e+003 iRT=33.4,33.4,33.4
NumPeaks: 31
308.1581 226.5 ?
309.0826 359.4 ?
325.6734 1596.8 b6-35^2/0.000,b6-36^2/0.492
329.1945 141.2 a6^2/0.000
343.1920 668.4 b6^2/0.000,y3-45/-0.042,y6-45^2/-0.516,y6-46^2/-0.024
353.2183 967.7 y3-35/0.000
354.2183 480.9 y3-35i/0.000
355.2183 481.6 y3-35i/0.000
358.1609 442.5 b3/0.000,y6-17^2/0.456,y6-18^2/0.948
358.1970 96.7 ?
365.2030 293.0 ?
373.2367 296.2 ?
388.2554 1744.6 y3/0.000,b7-36^2/-0.955
424.1714 315.8 b4-35/0.000,b4-36/0.984
442.1820 644.1 b4-17/0.000,b4-18/0.984
457.7633 546.4 p-45^2/0.000,p-46^2/0.492,y4-44/0.414
458.2791 109.5 p-44^2/0.000,b4/-0.929
471.7608 110.9 p-17^2/0.000,p-18^2/0.492
472.2608 111.1 p-17^2i/0.000
477.1321 111.6 ?
480.2740 336.0 p^2/0.000
484.3130 224.9 y4-17/0.000,y4-18/0.984
501.3395 523.5 y4/0.000
553.1674 240.4 ?
567.3501 487.3 y5-35/0.000
584.3766 247.1 y5-18/0.000
602.3872 2935.1 y5/0.000
696.3927 205.5 y6-35/0.000,y6-36/0.984
714.4032 2211.4 y6-17/0.000,y6-18/0.984
731.4298 10000.0 y6/0.000
732.4298 414.9 y6i/0.000

Name: DIETIIQK/2 (new entry, combined library)
LibID: 4673
MW: 960.5481
PrecursorMZ: 480.2740
Status: Normal
FullName: X.DIETIIQK.X/2 (CID-QTOF)
Comment: AvePrecursorMz=480.5611 BinaryFileOffset=20613807 FracUnassigned=0.09,1/5;0.25,13/20;0.17,13/31 MassDiff=-0.0014 Mods=0 NAA=8 NISTProtein=Id_02 NMC=0 NTT=1 Nreps=1/1 OrigMaxIntensity=11 Parent=480.274 Pep=Tryptic PepContext=1/NTPR_DIHE PrecursorIntensity=0 Prob=1.0000 Protein=1/Id_02 RawSpectrum=file_02_04.69683.69683 RetentionTime=2745.2,2745.2,2745.2 Sample=1/sample_02,1,1 Se=1^C1:pb=1.0000/0,fv=2.7040/0 Spec=CREATE TotalIonCurrent=8.9e+003 iRT=33.4,33.4,33.4
NumPeaks: 31
308.1581 226.5 ?
309.0826 359.4 ?
326.1680 1596.8 ?
328.2231 141.2 m4:6/0.000
344.1816 668.4 m3:5/0.000
353.2183 967.7 y3-35/0.000
354.2183 480.9 y3-35i/0.000
355.2183 481.6 y3-35i/0.000,m5:7/-0.016
358.1609 442.5 b3/0.000
358.1970 96.7 ?
365.2030 293.0 ?
373.2367 296.2 ?
388.2554 1744.6 y3/0.000
424.1714 315.8 b4-35/0.000
441.1980 644.1 b4-18/0.000
457.2713 546.4 p-46^2/0.000,m3:6/0.006
459.2086 109.5 b4/0.000
471.2688 110.9 p-18^2/0.000
472.2512 111.1 ?
477.1321 111.6 ?
480.2410 336.0 ?
484.3130 224.9 y4-17/0.000
501.3395 523.5 y4/0.000
553.1674 240.4 ?
568.3124 487.3 ?
584.2763 247.1 ?
602.3872 2935.1 y5/0.000
695.4087 205.5 y6-36/0.000
713.4192 2211.4 y6-18/0.000
731.4298 10000.0 y6/0.000
732.4675 414.9 ?

I assumed before, that it was an issue with the consensus building, but I figured out that it took action during the import via spectraST into MRM format.

Now, I just saw that I don't have this problem when I use the single libraries. Even with consensus libraries built with with an older spectraST version the MRM import works just fine.
Maybe there is the problem in the combination of the spectral libraries? I already got them as single consensus libraries and combined them with the following command

spectrast -cNcombined_lib -cd -cu -cr1 -cy1 -cAC -cJU Lib_01.splib Lib_01.splib Lib_01.splib Lib_01.splib

Could this be an issue here. Do I have to rebuild all libraries before combining them?

I would be really happy about any suggestions. Thank you in advance!

Kind regards,
Kristina
____________________
Kristina Plate, MSc. Biomathematics

University of Greifswald
Center for Functional Genomics of Microbes
Institute for Microbiology
Department of Microbial Proteomics
Felix-Hausdorff-Str. 8
17489 Greifswald

Eric Deutsch

unread,
Sep 20, 2019, 2:49:28 PM9/20/19
to spctools...@googlegroups.com, Eric Deutsch

Hi Kristina, I’m not certain I fully understand your issue, but maybe I can answer some questions:

 

Partial old:

Name: DIETIIQK/2 (old entry, original library)

308.1581  226.5     ?   

309.0826  359.4     ?   

325.6734  1596.8    b6-35^2/0.000,b6-36^2/0.492

329.1945  141.2     a6^2/0.000    

343.1920  668.4     b6^2/0.000,y3-45/-0.042,y6-45^2/-0.516,y6-46^2/-0.024   

 

Partial new:

Name: DIETIIQK/2 (new entry, combined library)

308.1581  226.5     ?   

309.0826  359.4     ?   

326.1680  1596.8    ?   

328.2231  141.2     m4:6/0.000    

344.1816  668.4     m3:5/0.000    

 

The first two peaks are fine. For the third entry, I am guessing that in the original library this peak got shifted to exact m/z of the presumed interpretation b6-35^2. But this is a highly unlikely interpretation. So in the new entry, that peak retains its original m/z (i.e. not snapped to a presumed interpretation) and a “?”. SpectraST seems to be smarter about not assigning a silly interpretation.

 

For the next peak with intensity 141.2, the old library had an interpretation of a6^2 and the m/z is snapped to that interpretation. But a6^2 is a super unlikely interpretation, or maybe let’s call it impossible and wrong. In the new library, this same peak has been interpreted as an internal fragmentation ion and the m/z has been snapped to that. The m4:6 means that the peptide was fragmented in two places such that you get a little ion that is TII and has a charge so you see it. Internal fragmentation turns out to be fairly common and so m4:6 is far more plausible an interpretation than a6^2. Same for m3:5 (ETI). Not necessarily correct, but far more plausible than an a6^2 and a b6^2.

 

So I’m guessing that the first spectrum comes from a version where internal fragmentation was not yet supported, and the second spectrum comes from a newer version of SpectraST that knows about internal fragmentation and interprets peaks better.

 

Does that seem like a reasonable explanation of what’s going on? Henry might know more.

 

Regards,

Eric

--
You received this message because you are subscribed to the Google Groups "spctools-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to spctools-discu...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/spctools-discuss/1573a2bc-1b4a-4ca7-bc88-75579028df95%40googlegroups.com.

Kristina Plate

unread,
Sep 23, 2019, 7:21:30 AM9/23/19
to spctools-discuss
Hi Eric,

thank you very much for your reply and the detailed and, yes, quite reasonable explanation. Some of the spectral libraries I got for the combination were built with SpectraST 4.0 and the remaining builds as well as the combination and further import steps were done with SpectraST 5.0. But there are two factors bothering me:

1. I thought too, that the "issue" is caused by different SpectraST versions. But when I did the MRM import (spectraST 5.0) on the single spectral libraries which were built with the older version (see above), this phenomenon didn't occur.

2. Even if it's explainable by a smarter interpretation by SpectraST, does that mean that I can use these changed spectra or is it questionable to use spectral libraries of older software versions imported with a newer one?

Thank you and kind regards,
Kristina

To unsubscribe from this group and stop receiving emails from it, send an email to spctools...@googlegroups.com.

Eric Deutsch

unread,
Sep 23, 2019, 11:48:10 AM9/23/19
to spctools...@googlegroups.com, Eric Deutsch

Hi Kristina, I am not so sure about answers to your questions. It would be ideal to use SpectraST 5.0 in all your processing if you can easily do this, yes, but in some cases it would not matter much. In general, where peak interpretation is involved, SpectraST 5.0 is better. But I don’t know relative to what you’re doing exactly when peak re-interpretation happens.

 

Can you just rebuild all your libraries with SpectraST 5.0? That would be safest.

To unsubscribe from this group and stop receiving emails from it, send an email to spctools-discu...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/spctools-discuss/d206982e-1892-494c-8762-0c519ae3b10c%40googlegroups.com.

Kristina Plate

unread,
Sep 24, 2019, 12:54:33 PM9/24/19
to spctools-discuss
Hi Eric,

I'm sorry, if it's a little bit confusing. Theoretically I can rebuild the spectral libraries but it will be difficult to get all pepxml files. So I just want to know if it is risky to take this mixed approach.

It seems to be a thing of the mix up of older and newer spectrast version, because if I built a spectral library myself fully with spectraST 5.0 there are no m ion annotations no matter which import parameters I use. So, even if these m ions are more plausible they don't occur if I built it from the scratch.

Also it's interesiting that these changes only occur during import from splib files, not from msp files. Further, libraries with m ion annotations change "back" to the m ion free annotation if they are imported from msp files.

(1) original:

Name: AAAALILAGLVADGK/2
301.1492 347.0 y6^2/-0.530,y3-17/-0.985,y3-18/-0.001 48/44 0.0057|0.58
311.1680 414.6 ? 50/44 0.0054|0.72
319.1597 1481.3 y3/-0.002 71/44 0.0025|0.32
327.2017 2003.3 ? 71/44 0.0048|0.30
341.2172 726.1 ? 59/44 0.0060|0.42


(2) after import (1) from splib file with -cM parameter:

Name: AAAALILAGLVADGK/2
301.1506 347.0 y3-18/0.000 48/44 0.0057|0.58
311.1680 414.6 ? 50/44 0.0054|0.72
319.1612 1481.3 y3/0.000 71/44 0.0025|0.32
327.2027 2003.3 m2:5/0.000 71/44 0.0048|0.30
341.2183 726.1 m9:12/0.000 59/44 0.0060|0.42


(3) after reimport (2) only from msp file (w or w/o -cM parameter):

Name: AAAALILAGLVADGK/2
301.6790 347.0 y6^2/0.000,y3-17/-0.456,y3-18/0.528 48/44 0.0057|0.58
311.1680 414.6 ? 50/44 0.0054|0.72
319.1612 1481.3 y3/0.000 71/44 0.0025|0.32
327.2027 2003.3 ? 71/44 0.0048|0.30
341.2183 726.1 ? 59/44 0.0060|0.42

All steps were done with the same version of spectraST. Especially the first line: where the software gets the information about the other alternative fragment types (the msp file was the only file in directory)?
I'm sorry if this is confusing but I just want to understand what is happening during the processes.

Thank you very much for your patience!
Kind regards,
Kristina

Eric Deutsch

unread,
Sep 24, 2019, 8:38:29 PM9/24/19
to spctools...@googlegroups.com, Eric Deutsch

Hi Kristina, I’m not fully following the steps and how they are related. Can you describe what “original” is and where it came from (incl. SpectraST version) and then provide the subsequent steps (actual commands) (including renaming to msp?)

 

Thanks,

To unsubscribe from this group and stop receiving emails from it, send an email to spctools-discu...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/spctools-discuss/36b5011c-8fe7-4bf0-b4ab-990ffe31eadd%40googlegroups.com.

Kristina Plate

unread,
Sep 26, 2019, 8:59:52 AM9/26/19
to spctools-discuss
Hi Eric,

the original spectral library is built from ipro.pep.xml files with SpectraST (version 5.0, TPP v0.0.0 TRUNK (DEV), Build 201408281759-6544:6594M (mingw-i686))). The spectral library sptxt file was manually changed (only small changes in protein name, Comment section) and reimported ((1) original):

spectrast -cNspectral_library -c_IRTiRT.txt -c_IRR -cfProtein !~ REVERSE_ -cr1 -cy1 -cAC -cJU -cP0.95 file01_interact.iproph.pep.xml file02_interact.iproph.pep.xml ... fileN_interact.iproph.pep.xml

spectrast -cNspectral_library02 spectral_library.sptxt.


This is the library I got for further processing. Now I simply imported this library to MRM format with SpectraST (version 5.0, TPP v5.1.0 Syzygy, Build 201711031215-7670 (Windows_NT-x86_64)) from splib file ((2) after import (1) from splib file with -cM parameter):

spectrast -cNspectral_library03 -cM -cICID-QTOF spectral_library02.splib.


The third step I did just for testing the influence of the file type ((3) after reimport (2) only from msp file (w or w/o -cM parameter)):

spectrast -cNspectral_library04 -cM -cICID-QTOF spectral_library03.sptxt.


The same phenomenon appeared when the original library was built with SpectraST (version 4.0, TPP v4.6 OCCUPY rev 0, Build 201208211847 (linux)). 

Does this help?
Thank you and kind regards,
Kristina

Reply all
Reply to author
Forward
0 new messages