I seem to loose all 2+ and 3+ peptides during Xinteract (TPP on Mac)

[Bjorn]

Hi,

I can do a tandem database search that results in a .tandem file.
I successfully convert it into a .tandem.pep.xml where I clearly see
that I have 2+ and 3+ spectra with hyperscores of +700.
When I run peptideprophet on that file, the result is a .pep.xml file
with only the 1+ left (no 2+ or 3+)
Where have all the good spectra I had before Xinteract gone??
I would expect those to be still there and have high probability
scores...

Any suggestion is welcome.

I'm using TPP 4.4.0 on Mac OS X 10.6.4.

Bjorn

[Bjorn]

and it has to be "lose", not "loose"...
My mistake.

[Bjorn]

Hi,
Can anyone shed some light on why the "Mixture model quality test
failed for charge (2+) and (3+)". They are the majority of my spectra
and I get good hyperscores especially from the 3+ spectra prior to
xinteract command. I do not feel my data is bad, but apparently
peptide prophet cannot make anything out of it...

I'm using TPP 4.4.0 on Mac (gui and command line), but I can reproduce
the error on the windows TPP 4.4.0. (also gui and command line).

Thanks,

Bjorn

PeptideProphet (TPP v4.4 VUVUZELA rev 0, Build 201008200839 (linux))
AKeller@ISB
read in 695 1+, 7797 2+, 9641 3+, 0 4+, 0 5+, 0 6+, and 0 7+ spectra.
Initialising statistical models ...
negmean = -0.94065
negmean = -1.41046
negmean = -1.64247
negmean = -0.2
negmean = -0.2
negmean = -0.2
negmean = -0.2
Iterations: .........10.........20....
WARNING: Mixture model quality test failed for charge (2+).
WARNING: Mixture model quality test failed for charge (3+).
model complete after 25 iterations
command completed in 3 sec

[Claire]

Hi Bjorn,

Did you ever get a solution to this problem?

I keep encountering the same issue.

Thanks,

Claire

[David Shteynberg]

Unfortunately your "good" spectra scores overlap significantly with you bad spectrum scores.  So PeptideProphet has trouble classifying this dataset.   Can you tell us more about your dataset and the search parameters used?  Does your database contain any decoys?

Thanks,
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/-/WFHXaitAt_4J.
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.

[Claire]

Hi David,

My dataset is of N-terminal dimethyl labelled peptides (also natural acetylated N-terminal peptides included in the mix..) analysed on an LTQ-Orbitrap. I am new searching against a database (Mouse IPI v3.87) containing decoys and have now been using the PeptideProphet option ' Only use Expect Score as the discriminant - helpful for data with homologous top hits, e.g. phospho or glyco (Tandem only) '. Both of these have improved my models for +3 and +4 but the model for +2 is still poor.

I've had the same dataset analysed on LTQ-Orbitrap using ETD for >+3 charge and modelling in TPP looks great for >+3 charge data but is still poor for +2 charge.

I've attached example of X!Tandem parameter file that I'm currently using for searches.

Any suggestions on how I can improve modelling for +2 charge would be great.

Thanks,

Claire