Xilmass: A New Approach toward the Identification of Cross-Linked Peptides

Chemical
cross-linking coupled with mass spectrometry plays an
important role in unravelling protein interactions, especially weak
and transient ones. Moreover, cross-linking complements several structural
determination approaches such as cryo-EM. Although several computational
approaches are available for the annotation of spectra obtained from
cross-linked peptides, there remains room for improvement. Here, we
present Xilmass, a novel algorithm to identify cross-linked peptides
that introduces two new concepts: (i) the cross-linked peptides are
represented in the search database such that the cross-linking sites
are explicitly encoded, and (ii) the scoring function derived from
the Andromeda algorithm was adapted to score against a theoretical
tandem mass spectrometry (MS/MS) spectrum that contains the peaks
from all possible fragment ions of a cross-linked peptide pair. The
performance of Xilmass was evaluated against the recently published
Kojak and the popular pLink algorithms on a calmodulin-plectin complex
data set, as well as three additional, published data sets. The results
show that Xilmass typically had the highest number of identified distinct
cross-linked sites and also the highest number of predicted cross-linked
sites.