Comparing sage with spectronaut in DIA data

[pcarvalho75]

Hey guys,

I decided to take sage for a spin in single cell proteomic data. To achieve this I compared my results with those obtained from spectronaut. The search comprised a 20mz windows. While Im getting around 500 peptides with spectronaut, these map to roughly 250 proteins which is nice because i get proteins with 2 peptides, and no reverses were reported. For sage I get about 500 peptides but half the proteins are reverses, showing the results are are just by chance. In this regard, i believe to be doing something terribly wrong and would greatly appreciate any assistance. I converted my raw files using msconvert with default parameters. Any help would be more than welcome. I made available a folder with the raw file, my spectronaut results and my sage results here: https://1drv.ms/f/c/dc8e744a061b369b/EmTFUy-b2zlDmGTuYuSDSeEB7Cyu-hYmSifsAOAX9Adr-w?e=8ML8lF

[lazear]

Hi, depending on window width you may want to change the report_psms parameter. Sage is a spectrum-centric DDA search engine - while it can work for DIA data, it will almost always perform worse than dedicated DIA software. The disparity will be magnified by the quality/abundance/isolation-window-width of the DIA data. If you have lots of peptide detections relying on < 5 fragments, then it's going to be pretty difficult for Sage to discriminate between decoys and target hits, since a spectral library/known RTs aren't taken into account.