Quantitative mass spectrometry-based proteomic experiments are designed to answer increasingly complex questions. For example, we may want to characterize changes in protein abundance in biological samples collected repeatedly over time and across multiple conditions. Or we may want to distinguish changes in protein abundance from other changes, such as changes in post-translational modifications. At the same time, proteomics experiments increasingly incorporate Tandem Mass Tags (TMT) labeling, a multiplexing strategy that gains both accuracy of relative protein quantification and sample throughput. Combining complex designs and TMT multiplexing leads to unique interplays of multiple sources of systematic variation of interest, and of nuisance variation. This talk will advocate for multilevel modeling for statistical analyses of data in such complex situations, and its implementation in the open-source R/Bioconductor family of MSstats packages, specifically MSstatsTMT, MSstatsPTM and MSstatsShiny.