#36 · Proteomics

Computational protein localization and secretory pathway inference from proteomic data

The dominant thread is computational inference of protein location and functional organization from proteomic data. FSPmix, an open-source R package, applies semi-supervised functional clustering to spatial proteomics to predict protein subcellular localization. secRecon takes a different route, offering a curated reconstruction of 1,127 mammalian secretory pathway genes across 77 process terms, applied to plasma cell and CHO omics data to find mechanisms regulating antibody production. A third preprint, which benchmarks six copula models of gene coexpression in scRNA-seq data, is less directly connected to proteomics; it reports that Gaussian copulas balance accuracy and speed while vine copulas scale poorly, extending the modeling framing to transcriptomic dependence structure. What is new is the combination of structured pathway reconstruction applied to secretory biology with spatial localization inference delivered in open-source form.