Peptide Correlation Analysis (PeCorA) Reveals Differential Proteoform Regulation

Shotgun
proteomics techniques infer the presence and quantity of
proteins using peptide proxies produced by cleavage of the proteome
with a protease. Most protein quantitation strategies assume that
multiple peptides derived from a protein will behave quantitatively
similar across treatment groups, but this assumption may be false
due to (1) heterogeneous proteoforms and (2) technical artifacts.
Here we describe a strategy called peptide correlation analysis (PeCorA)
that detects quantitative disagreements between peptides mapped to
the same protein. PeCorA fits linear models to assess whether a peptide’s
change across treatment groups differs from all other peptides assigned
to the same protein. PeCorA revealed that ∼15% of proteins
in a mouse microglia stress data set contain at least one discordant
peptide. Inspection of the discordant peptides shows the utility of
PeCorA for the direct and indirect detection of regulated post-translational
modifications (PTMs) and also for the discovery of poorly quantified
peptides. The exclusion of poorly quantified peptides before protein
quantity summarization decreased false-positives in a benchmark data
set. Finally, PeCorA suggests that the inactive isoform of prothrombin,
a coagulation cascade protease, is more abundant in plasma from COVID-19
patients relative to non-COVID-19 controls. PeCorA is freely available
as an R package that works with arbitrary tables of quantified peptides.