Param-Medic: A Tool for Improving MS/MS Database Search Yield by Optimizing Parameter Settings

Abstract

In shotgun proteomics analysis, user-specified parameters are critical to database search performance and therefore to the yield of confident peptide-spectrum matches (PSMs). Two of the most important parameters are related to the accuracy of the mass spectrometer. Precursor mass tolerance defines the peptide candidates considered for each spectrum. Fragment mass tolerance or bin size determines how close observed and theoretical fragments must be in order to be considered a match. For either of these two parameters, too wide a setting yields randomly high-scoring false PSMs, whereas too narrow a setting erroneously excludes true PSMs., in either case lowering the yield of peptides detected at a given false discovery rate. We describe a strategy for inferring optimal search parameters by assembling and analyzing pairs of spectra that are likely to have been generated by the same peptide ion in order to infer precursor and fragment mass error. This strategy does not rely on a database search, making it usable in a wide variety of settings. In our experiments, this strategy yields more high confidence PSMs than using settings based on instrument defaults or determined by experts. Param-Medic is open source and cross-platform. It is available as a standalone tool (http://noble.gs.washington.edu/proj/param-medic/) and has been integrated into the Crux proteomics toolkit (http://crux.ms), providing automatic parameter selection for the Comet and Tide search engines.

Improving power while controlling the false discovery rate when only a subset of peptides are relevant

“…Database searches were conducted using Tide with the combined p-value score function 23 against a concatenated target-decoy database. The precursor mass tolerances, estimated by Param-Medic, 24 were found to be 31 ppm for the concatenated UPS1/yeast runs, 80 ppm for the castor plant runs, 63 ppm for the previously published UPS1/yeast runs, and 38 ppm for the human runs. All Tide parameters were set to their default values, except an isotope error of 1 was allowed and the "top-match" parameter (i.e., number of reported PSMs per spectrum) was set to 10,000.…”

More Accurate Transcript Assembly via Parameter Advising

“…In contrast, separate work has been done in other fields on a priori selection, where the parameters are chosen in advance by looking at the raw input alone, or a subsample of the input along with its result. This includes work such as Param-Medic (May et al, 2017 ) for choosing mass-spectrometry database search parameters, KmerGenie (Chikhi and Medvedev, 2013 ) for finding appropriate k -mer sizes for genomic assembly, and GRAPE (Majoros and Salzberg, 2004 ), which finds model parameters for gene finding. There is also work on this problem outside of computational biology such as ParamILS (Hutter et al, 2009 ), which finds optimal settings for the CPLEX computational optimization tool, SATZilla (Xu et al, 2008 ) for choosing from a collection of SAT solvers, as well as many tools developed for tuning hyperparameters in machine learning such as TPOT (Olson et al, 2016 ), which uses genetic algorithms, and Spearmint (Snoek et al, 2012 ), which uses Bayesian optimization.…”

mokapot: Fast and flexible semi-supervised learning for peptide detection

“…Additionally, carbamidomethylation of cysteine was specified as a static modification. We selected the precursor m/z window using Param-Medic [24] and set a fragment ion tolerance of 0.02 Da. The protein database was processed using Tide to generate a shuffled decoy peptide sequence for each peptide sequence in the target database, preserving both termini.…”

“…The protein database was processed using Tide to generate a shuffled decoy peptide sequence for each peptide sequence in the target database, preserving both termini. We selected the precursor m/z window using Param-Medic [24] and set a fragment ion tolerance of 0.02 Da. We included the TMT 10-plex modification of lysine and the peptide N-terminus as a static modification, but carbamidomethylation of cysteine was not included.…”