spqrp
Detect sample-provenance inconsistencies and potential mix-ups in mass-spectrometry-based plasma-proteome cohorts. Provides a clustering-based approach (build a nearest-neighbour graph in a dimensionality-reduced space and iteratively split large components by edge weight), a threshold-based approach (classify sample pairs as belonging or not-belonging from a pairwise distance cutoff), parameter optimization over distance metrics and cutoffs, and a pairwise random-forest classifier for protein importance ranking. This is a native R port of the author's Python package 'spqrp' (<https://github.com/fhradilak/spqrp>), implementing methods from an associated manuscript currently in preparation.
This page was built for software: spqrp
