We are doing mass spectrometry on (semi-)purified protein samples. The aim is to reach a maximum amount of sequence coverage. Since we use a purified mixture, software like percolator eliminates too much positive hits.
Before, we used Mascot with an ionscore cutoff of 20 (consistent with a 1% chance the match is a False Positive). Now, we have switched to Sequest and I am having difficulty finding the correct Xcorr cutoff which accomplishes the same thing. I hear I should use different cutoffs for different charge states but there is little to no consistency in which value to chose.
Can someone help me to find the best cutoff for these experiments?
Alexander,
the FDR at a certain score cutoff will vary with the TDA. To get the FDR as a function of the search engine (Sequest) score, you could use a large synthetic peptide library to derive a model (see publication).
A more general approach would be to filter the data at a certain FDR but relax the cutoff to 0.05 or 0.10.
HTH
Harald
Article A large synthetic peptide and phosphopeptide reference libra...
Harald, thank you for your answer. Relaxing the cutoff is of course something we thought of before, but then again comes the question: how far can you go with relaxing the cutoff.
The problem is that even very poorly scored PSM's sometimes have quite decent spectra with multiple matching peaks and low mass differences. So that's why we wanted to find a different method than Percolator.
I will look into your publication as well. Thanks for sharing!
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics