Scaffold 4.11 comes with significant changes to how Proteome Discoverer data is handled. We have noticed that some datasets searched with Sequest HT resulted in very high FDR values when loaded into Scaffold. This is due to the interaction of Sequest data with Scaffold's internal LFDR probability assignment algorithm. Note, this does not apply to data that was searched with Mascot through Proteome Discoverer. For those datasets, LFDR handles the results quite well.
It has previously been our recommendation to use the Target/Decoy PSM validator node in the processing workflow and to set the strict FDR, relaxed FDR and deltCN to 1. This allowed for all of the PSMs to be written to the MSF file and provided more data for Scaffold's probability model. This resulted in more accurate probability assignments.
However, with the release of Scaffold 4.11 we are now recommending using the Percolator node in the processing workflow to control FDR. Keep the FDR set to an appropriate value when processing data. In most cases this will be 1%.
Then load the data into Scaffold using prefiltered mode. This will bypass the probability model in Scaffold. Note, if you want to use prefiltered mode, make sure you disable the option to use condensing when loading data.
In previous versions of Scaffold, when you loaded data using prefiltered mode you would not be able to threshold you data using the dropdown menus in the Samples view. In Scaffold 4.11 we have added the ability to threshold your data using the probabilities that were assigned by Percolator. Thus not only will the probability assignments be more reliable but you will now be able to use them for thresholding purposes. If you are using a pre-4.11 version of Scaffold, we suggest applying a custom peptide threshold using the Edit menu to PD results to filter out lower scoring matches after loading with prefiltered mode.
We recommend anyone using Proteome Discoverer update to Scaffold 4.11. This is particularly important for those who often use Sequest HT. This workflow will work for those using Mascot as part of PD as well. Note, the other probability assignment options will still be available in Scaffold and can be used when needed. That being said we feel in most cases that you will get the best results from Sequest HT data if you follow this new procedure.
An example processing workflow for precursor intensity data from PD 2.4. Note, the Percolator node is used for probability assignments.
Clicking on the Percolator node displays the above parameters. Keep the deltaCN and FDR targets set to an acceptable value for later processing in Scaffold. Because the relaxed FDR is set to 0.05 here you will accept peptides up to a 5% FDR.