X!TandemPipeline++: Software for Ion Mobility-Enabled Quantitative Proteomics in timsTOF Data Format
Résumé
Introduction
X!TandemPipeline (Langella et al. 2017) is a proteomics free and open source
Java software program designed to filter and group peptide/protein
identifications from MS/MS mass spectra. After a complete rewrite in C++17,
X!TandemPipeline++ now features both native support for the timsTOF raw data
format and peptide/protein quantification. X!TandemPipeline++ performs peptide
identifications and area under the curve XIC-based quantifications using the
Bruker's native raw data format. Using a common HeLa data set published by
Meier et al. (2018, PXDO1OO12), we demonstrate that X!TandemPipeline++
identifies and quantifies significantly more proteins than competitors MaxQuant
and MSFragger. It is also significantly faster.
Methods
X!TandemPipeline++ is written in portable C++17 and makes use of the Qt
libraries for the graphical user interface. Binary packages are available for
Linux and MS Windows. The timsTOF native raw data reader was developed in-house with
the technical specifications provided by Bruker. The sofware was
tightly optimized to ensure very fast access to the binary data. The current
version provides real time MS/MS peptide annotation and performs extremely fast
ion current extractions and XIC chromatogram visualizations (typically, in less
than five seconds for a 2 hour PASEF run).
Preliminary Data or Plenary Speakers Abstract
The Bruker timsTOF line of instruments improves the identification of peptides
and proteins in complex mixtures by implementing a peculiar ion mobility
technology. X!TandemPipeline++ has the distinct feature, with respect to the
MaxQuant and MSFragger competitors, of natively parsing the timsTOF raw data
with original software code that puts us in total control of the data
processing, in particular where tight optimizations are needed for speed or in
cases where special data processing features need adding for increased accuracy.
This is exemplified by X!TandemPipeline++ using the renowned X!Tandem engine
(Craig et al. 2004) to perform database searches right on optimized data
brokered to it by our reader of native timsTOF binary data.
Our software identifies roughly the same amount of peptides as when using the
X!Tandem engine on MGF data provided by the Bruker Data analysis tool but
performs much faster, reducing processing time from 55 to 12 minutes on a
typical quality control HeLa sample. For peptide quantifications,
X!TandemPipeline++ uses the ion mobility-enabled version of MassChroQ (Valot et
al. 2011), both software pieces sharing our optimized native timsTOF data
reader.
Using a common HeLa data set (Meier et al. (2018), PXDO1OO12, 4 technical
replicates), we found more quantified proteins with a minimum of 2 quantified
ions : 5783 vs 4977 (MSFragger), 3526 (MaxQuant). Correlation of quantified
proteins between samples was 0.99. Median protein coefficient of variation (CV)
was also very good : 0.059 vs 0.049 (MSFragger), 0.072 (Peaks) and 0.070
(MaxQuant). Comparable results have been found using data set PXD014777
(Prianichnikov et al., 2020, a mixture of proteins from three organisms, H.
sapiens, S. cerevisiae, and E. coli). Overall, while producing similar or
better results than competing software, X!TandemPipeline++ always performs
faster than all the other software offerings.
Novel Aspect
Fast and accurate timsTOF native raw data support for quantitative proteomics.
Origine : Fichiers produits par l'(les) auteur(s)