Only triply charged precursors!!!!

[Alex]

Hi TPP community!

I have an interesting puzzle that I can not figure out!!! I'm running a 2D-LC-MS/MS experiment of a yeast lysate (peptides labeled by dimethylation) on a QqLIT (QTrap 3000) instrument acquiring on IDA mode.

I'm analyzing the data on the TPP using X!Tandem search engine with the params file attached, and interestingly, ALL the peptides identified on the pep.shtml file are derived from triply charged precursors!!! (ALL of them). When I check the prot.shtml file all the identified peptides have the 3_SEQUENCE nomenclature (my understanding is that the 3 identifies the charge on the precursor, right?). When I look at the raw data (.wiff) file on Analyst software, I can easily estimate that most (probably a little more than half) of the MS/MS was actually performed on CLEARLY doubly charged precursors, according the the ER-MS (enhanced resolution MS acquired on the LIT)... so it seem very weird that I dont find a single peptide to spectrum match (PSM) from a doubly charged precursor. Just wondering if I might have accidentally changed a parameter on the params file (.xml).

Has anybody crossed upon this observation before??? Anybody can help me???

Thanks!!!

Alex

[Alex]

I just wanted to add that I checked on previously analyzed samples, and that apparently, with the same settings and experimental conditions, I got the expected distribution of 2+ and 3+ peptides before. This sample was acquired on march of this year, probably with an earlier version of TPP. After that acquisition, I checked my results, and I have only identified 2+ OR 3+ peptides in the pep.shtml files (I didn't realize back then on this peculiar observation). Could someone check if this has happened in your set of results?

Thanks!

Alex

[Alex]

Another update to this situation!

I think the problem might actually be in PeptideProphet. I run the same dataset using X!Tandem on the GPM.org website, and I do identify doubly charged precursor peptide ions. The same dataset only shows triple charged precursors, as observed in the pep.shtml file. I select the column: z (assumed charge), and ONLY triple charged species appear. 

Here is another twist: When I go to Peptide Prophet Analysis results I only see models for 3+, and 

No models or data found for charge +2.

However if I hit on show ALL models, I get the following (which I'm not an expert to understand, so hopefully someone can fill me on this information):

FINAL 2+ MODEL after 24 iterations: number of spectra: 53628 using 3+ positive distributions to identify candidates ('-2') above background ('0') prior: 0.000, total: 0.0 X! Tandem (k-score) discrim score [fval] negmean: -1.33 pos: (gaussian mean 1.55, stdev 2.01) neg: (evd mean -1.31, stdev 1.14, mu -1.82, beta 0.89) no. tolerable trypsin term. [ntt] pos: (ntt=0 0.000, ntt=1 0.000, ntt=2 1.000) neg: (ntt=0 0.000, ntt=1 0.000, ntt=2 1.000) no. missed enz. cleavages [nmc] pos: (nmc=0 0.878, 1<=nmc<=2 0.122, nmc>=3 0.000) neg: (nmc=0 0.409, 1<=nmc<=2 0.591, nmc>=3 0.000) accurate mass diff [massd] (offset: -0.500000) pos: (massd=-5.000000 0.000000000, massd=-4.000000 0.000000000, massd=-3.000000 0.000000000, massd=-2.000000 0.000154395, massd=-1.000000 0.029449633, massd=0.000000 0.390222698, massd=1.000000 0.422314370, massd=2.000000 0.128208899, massd=3.000000 0.017069506, massd=4.000000 0.012580499, massd=5.000000 0.000000000) neg: (massd=-5.000000 0.000000000, massd=-4.000000 0.000000000, massd=-3.000000 0.000000000, massd=-2.000000 0.000173670, massd=-1.000000 0.179740500, massd=0.000000 0.175030116, massd=1.000000 0.164730014, massd=2.000000 0.158361143, massd=3.000000 0.159298718, massd=4.000000 0.162665838, massd=5.000000 0.000000000)

FINAL 3+ MODEL after 30 iterations: number of spectra: 30124 using training data negative distributions prior: 0.023, total: 699.6 X! Tandem (k-score) discrim score [fval] negmean: -1.23 pos: (gaussian mean 5.15, stdev 2.54) neg: (evd mean -1.23, stdev 1.26, mu -1.80, beta 0.99) no. tolerable trypsin term. [ntt] pos: (ntt=0 0.000, ntt=1 0.000, ntt=2 1.000) neg: (ntt=0 0.000, ntt=1 0.000, ntt=2 1.000) no. missed enz. cleavages [nmc] pos: (nmc=0 0.878, 1<=nmc<=2 0.122, nmc>=3 0.000) neg: (nmc=0 0.409, 1<=nmc<=2 0.591, nmc>=3 0.000) accurate mass diff [massd] (offset: -0.500000) pos: (massd=-5.000000 0.000000000, massd=-4.000000 0.000000000, massd=-3.000000 0.000000000, massd=-2.000000 0.000154395, massd=-1.000000 0.029449633, massd=0.000000 0.390222698, massd=1.000000 0.422314370, massd=2.000000 0.128208899, massd=3.000000 0.017069506, massd=4.000000 0.012580499, massd=5.000000 0.000000000) neg: (massd=-5.000000 0.000000000, massd=-4.000000 0.000000000, massd=-3.000000 0.000000000, massd=-2.000000 0.000173670, massd=-1.000000 0.179740500, massd=0.000000 0.175030116, massd=1.000000 0.164730014, massd=2.000000 0.158361143, massd=3.000000 0.159298718, massd=4.000000 0.162665838, massd=5.000000 0.000000000)

Hope this helps to figure out why I dont get any information for doubly charge precursors, which are clearly there on my .wiff data!

Thanks again and hope someone will pick on with this!!!

Alex

[Alex]

to complete the picture, here the proteinprophet analysis results

ProteinProphet® analysis results
Version:  Insilicos_LabKey_C++ (TPP v4.5 RAPTURE rev 2, Build 201202031108 (MinGW))
Analysis Date: 2012-07-04T12:21:56

Source Files: c:/Inetpub/wwwroot/ISB/data/120629YEAST2D/interact.pep.xml

Number of input spectra with minimum probability 0.05: 0 1+, 0 2+, and 1324 3+

Reference Database: c:/Inetpub/wwwroot/ISB/data/dbase/Yeast/YEAST.fasta
Residue substitutions: I -> L
Run Options: 
	Occam's Razor used: Y
	Protein Groups: Y
	Peptide degeneracies: Y
	Number of Sibling Peptides used: Y
	Min peptide probability: 0.20
	Min peptide weight: 0.50

Original results written to file: c:/Inetpub/wwwroot/ISB/data/120629YEAST2D/interact.prot.xml


Analysis Iterations
EM step
number of iterations
Initial Peptide Weights
2
NSP Distributions
2
Final Peptide Weights
1


Learned Number of Sibling Peptide Distributions
Neighboring bin smoothing: Y
bin number
nsp range
positive freq
negative freq
positive/negative ratio
0
0.00 <= nsp < 0.10
0.208
0.535
0.39
1
0.10 <= nsp < 0.25
0.065
0.142
0.46
2
0.25 <= nsp < 0.50
0.058
0.042
1.38
3
0.50 <= nsp < 1.00
0.124
0.074
1.67
4
1.00 <= nsp < 2.00
0.125
0.063
1.97
5
2.00 <= nsp < 5.00
0.166
0.059
2.82
6
5.00 <= nsp < 15.00
0.200
0.066
3.05
7
15.00 <= nsp < inf
0.054
0.019
2.81 (3.05)

      

Total Number of Contributing Spectrum IDs (100% share): 660.1

[David Shteynberg]

Hi Alex,

It sounds like perhaps the 2+ model failed the QC check and you got 0 probabilities for all 2+ PSMs.  I am not sure why this would be, one possibility is variable modifications that generate homologous top matches. You can try toaddress this with LEAVE or EXPECTSCORE options that ignore the scores from homologous matches. Also, you may try a decoy search and DECOY= and/or NONPARAM options which are helpful at separating the negative model from the positive model when the two overlap.

-David  

--
You received this message because you are subscribed to the Google Groups "spctools-discuss" group.

To view this discussion on the web visit https://groups.google.com/d/msg/spctools-discuss/-/wiv8gSP7xLgJ.

To post to this group, send email to spctools...@googlegroups.com.
To unsubscribe from this group, send email to spctools-discu...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/spctools-discuss?hl=en.