The Good, the Bad and the Ugly: when statistics tells you to throw away peptide IDs

[Orbitrap_SciLib]

Siegfried Gessulat1, Tobias Schmidt2, Michael Graber1, Florian Seefried1, Dave Horn3, Christoph Henrich4, BernardDelanghe4, Daniel Zolg2, Mathias Wilhelm2, Bernhard Kuster2, Martin Frejno1
ASMS 2020
Purpose: Identify the optimal PSM validation method available. Methods: Comparison of different PSM validation methods and their performance in separating targets from decoys. Results: Semi-supervised machine learning using multiple scores can separate targets from decoys better than classical approaches based on a single score but there is room for improvement.


1msAId GmbH, Garching, Germany; 2Technical University of Munich, Freising, Germany; 3Thermo Fisher, San Jose, CA; 4Thermo Fisher Scientific (Bremen) GmbH, Bremen, Germany