False discovery rate and mascot score for publishing

[neil777]

Hi
Does anyone know what is an acceptable false discovery rate for publishing in proteomics journals. Also is there an acceptable cutoff for mascot score, or can you include any id as long as the FDR is satisfied. Does anyone know the best way to decrease your FDR, for example if i narrow the mass tolerance on search parameters does that decrease identification of peptides belonging to the decoy database.
i appreciate any help
best wishes
neil

[Craig]

An FDR of 1% is pretty standard, but as long as you report what your FDR was you should be able to set whatever threshold you want within reason. There is no minimum Mascot score as long as you have appropriate FDR, although some questions have been raised about accepting very poor MS/MS spectra.

There are many strategies for decreasing your FDR. Narrowing the mass tolerance is certainly one approach that can be effective. But it requires that you are discarding a high ratio of decoys to targets, which may not necessarily be the case, for example if you are already at the optimal mass tolerance. Any improvement of search parameters will be a similar story. You could also try incorporating additional criteria like secondary scores (e.g. deltaCn for Sequest) or retention time error, although this can require significant effort.

[aky]

You can also try Mascot percolator or peptide prophet for improving the results.

[karthikskamath]

Well said Craig!

This has been surely a topic of debate from a long while. This talk from Alexey Nesvizhskii provides some insights;

"Peptide ID and protein inference using PeptideProphet and ProteinProphet"

Also 2008 January's addition of Journal of Proteome research published many articles related this matter.