Prediction and Consequences of Cofragmentation in Metaproteomics

Metaproteomics
can provide critical information about biological
systems, but peptides are found within a complex background of other
peptides. This complex background can change across samples, in some
cases drastically. Cofragmentation, the coelution of peptides with
similar mass to charge ratios, is one factor that influences which
peptides are identified in an LC–MS/MS experiment: it is dependent
on the nature and complexity of this dynamic background. Metaproteomics
applications are particularly susceptible to cofragmentation-induced
bias; they have vast protein sequence diversity and the abundance
of those proteins can span many orders of magnitude. We have developed
a mechanistic model that determines the number of potentially cofragmenting
peptides in a given sample (called cobia, https://github.com/bertrand-lab/cobia). We then used previously published data sets to validate our model,
showing that the resulting peptide-specific score reflects the cofragmentation
“risk” of peptides. Using an Antarctic sea ice edge
metatranscriptome case study, we found that more rare taxonomic and
functional groups are associated with higher cofragmentation bias.
We also demonstrate how cofragmentation scores can be used to guide
the selection of protein- or peptide-based biomarkers. We illustrate
potential consequences of cofragmentation for multiple metaproteomic
approaches, and suggest practical paths forward to cope with cofragmentation-induced
bias.