Mascot Results --> imported in Scaffold -> does it consider Rank 2 or Rank 3 identifications from Mascot?

[jonas]

Hello everybody

We have the following request with Scaffold 2.

We imported some Mascot results into Scaffold and it seems that
Scaffold NEVER considers the second rank and even assigns a very high
probability to peptides
where the second ranked peptide is only marginally apart from the
first rank. To my understanding of PeptideProphet on the top of
Sequest results it would definitely decrease the confidence of
peptides even for very high XCorrs when the difference in XCorr
between the first and second best peptide is very small (delta score
is smaller than 0.1)!!

This scenario is of special interest in a Phospho-Proteomics approach
where we tend to differentiate between identified Phospho-peptide and
confident identified Phospho-site.
The difference is when we get a Phospho-peptide on rank 1 and the
score for rank 2 of the Mascot result is way below the score of rank
1.

This can be a problem for peptides which e.g. have several possible
phosphorylatable amino acids... (if a peptide contains more than one
Ser)
In such cases the score for either one of the two phosphorylated
Serines is about the same range or even exactly the same!!
---> (we would consider one identifications as definitely superior
over the other if the delta score is bigger than 19!)
---> if the delta score is not more than 19 apart we consider the
peptide as perfectly identified phospho-peptide but the phospho-site
cannot be matched!!

Is there any possibility to display the second, third, fourth .....
best peptide and its scores in Scaffold??

Regards,

jonas


-----******------*******-------********-----
Jonas Grossmann, PhD
Proteome Informatic Expert
Functional Genomics Center Zurich
FGCZ, ETH Zurich
Winterthurerstrasse 190
8057 Zurich, Switzerland

Tel: 0041 (0)44 635 39 10

[mpitman]

Hi Jonas,

Thanks for the post.

I know that we do consider the differently ranked peptides, but I am
not the best person to answer in detail. So, I'm cc'ing Brian Searle
who should be able to explain better how the algorithm considers the
different rankings.

Thanks,

Mark P.

[Brian Searle]

Hi Jonas,

That's a very good question! Scaffold has the ability to consider the second rank (up to five), but
there has to be significant evidence that it exists in the sample over the first ranked peptide
(i.e. another search engine agrees with the second rank). This method is discussed briefly in J.
Proteome Res. 2008; 7(1); 245-253. If there is no such evidence, then the first rank is chosen (it
has the most marginal evidence).

I agree that it would be a nice feature to display lower ranked peptide identifications,
particularly in the case of identifying modification sites where there is marginal localization
evidence. As of right now, there's no way to do that in Scaffold. Right now, the best way you can
improve the localization of modifications based off your spectral evidence is by using multiple
search engines that consider different aspects of each spectrum.

As a side note, the discriminant score that Peptide Prophet uses for Mascot is the ion score minus
the identity score. It doesn't take into consideration a deltaCn like term because the identity
score is generally a better estimate of the relative goodness of a score. For example, you could
imagine a situation where you identify two peptides PEPTIDER and EPPTIDER with similar scores
because the sequences are so similar, say ion scores of 65 and 64 with an identity score of 40.
However, you have fairly good evidence that one of those peptides exists in your sample because the
ion scores are so much higher than the identity score.

Best wishes,
Brian