1. David L. Tabb, Ph.D. January 11, 2017
Broadening the mission: quality assessment for quantitative and label-dependent mass spectrometry
David L. Tabb, Ph.D.
January 11, 2017

2. Organization Introducing the HUPO-PSI QC Working Group
Assessing isobaric labeling: Spectrum Mill
Understanding U.S. Pharmacopeia Chapter 621
Assessing quality in targeted expts: SProCoP/MSstats
Extending into MSE and SWATH
Planning for MALDI profiling and imaging

3. HUPO-PSI QC Working Group
Weimin Zhu Stefan Tenzer Wout Bittremieux Dave Tabb Mathias Walzer (plus not pictured) Martin Eisenacher Reza Salek
Produce format, CV terms, and infrastructure to analyze, visualize, and convey quality information

4. QC infrastructure serves many purposes
Figure by Wout Bittremieux

5. Spectrum Mill innovations
Tag-based searching
Ion series-based scoring
Raw data preprocessing
Homology searching
AA substitution
Wildcard PTMs
QC metric producing
Per-MS/MS assessing

6. iTRAQ labeling
After digestion, amino groups of peptides can be attached to isobaric tagging reagents.
Tags fragment in MS/MS to produce reporter ions, depending on tag type.
AB Sciex sells 4- and 8-channel reagents.
Ross et al (2004) Mol. Cell. Proteomics 3:

7. Why do isobaric tags introduce variability?
iTRAQ reagents bond to primary amines both at N-terminus and on lysine.
Lab errors may alter label rxn efficiency.
Fold changes are biased toward 1:1 ratios.
D.W. Mahoney et al. J. Proteome Res. (2011) 10:

8. Spectrum Mill 6.0 isobaric QC
For what fraction of MS/MS scans is label observed at N-term? At Lys? Neither? Both?
N-term is typically less than complete
What is the ratio of intensity for signal at a given reporter to most intense (across all MS/MS)?
What was median signal/noise ratio for reporter ion intensities?
Insights from Karl Clauser, Broad Institute

9. USP*: Useful Starting Point
*Actually Pharmacopeia of the United States of America, 38th Rev.

10. USP Chapter 621 on Chromatography defines:
Resolution
Separation Factor
Symmetry Factor
Signal-to-Noise Ratio
%RSD Requirements

11. Tools for SRM/XIC evaluation
SProCoP (Statistical Process Control in Proteomics): Bereman et al. JASMS (2014) 25:
MSstats: Choi et al. Bioinformatics (2014) 30:

12. SProCoP: key QC visualizations
Shewhart Process Control Chart
Pareto Chart Variability contribution
+2SD
+1SD
Mean
-1SD
-2SD
Intervention
Bereman et al. JASMS (2014) 25:

13. MSstats v3.7.1 differentiates, finds LOD, and monitors QC
DDA SRM DIA
MSstats can accommodate intensity data from any of these experiment designs, using peptide XICs (DDA) or peptide fragment XICs (SRM,DIA).
Visualization of one representative protein in a DDA, an SRM and a DIA experiment. Colors and shapes represent peptides, and multiple line types of a same color and shape represent the fragments of the peptide. Vertical lines separate times or conditions. (a) Protein alcohol dehydrogenase-yeast spiked into a complex background in six concentrations. (b) Protein ACH1, at 10 times points after a stress. (c) Protein FabG of Streptococcus, with 0 and 10% human plasma added
Choi et al. Bioinformatics (2014) 30:

14. Specializing QC metrics to DIA
Is cycle time a function of TIC?
Dot products can compare successive scans for a swath.
Mass precision can be examined through repeat scans.
Gillet et al. Mol. Cell. Proteomics (2012) /mcp.O –3

15. Opportunities in DIA precursor-fragment correlation
What fraction of MS isotope packets match fragments?
What is median RT diff for peak maxima?
How close to an exponential-Gaussian is elution profile?
Retention Time
Geromanos et al. Proteomics (2009) 9: Broeckling et al. Metabolomics (2012) 9:33
Kalambet et al. Chemometrics (2011) 25: 352

16. QC for MALDI Profiling
Within-cohort variability can be assessed on both m/z and intensity.
Matrix-associated peaks may reveal spot- dependent effects.
m/z mapping benefits from mass accuracy more than resolution.

17. QC for MALDI Imaging Spatial coords imply intensity similarity.
Is intensity “noise” independent of m/z?
TIC histogram may highlight regions of poor signal.

18. Closing Thoughts
Proteomic paradigms beyond DDA and SRM need more support in QC toolsets.
Sometimes essential work can appear less glorious, at first glance. That does not make it nonessential.
The HUPO-PSI QC Working Group is looking for brains and hands!