Generation of ground-truth data for the systematic evaluation of data normalization strategies in single-cell proteomics data

Single-cell technologies have become a fundamental part of biomedical research in the past years, yet these have been mainly limited to the study of transcriptomes. In contrast, proteomic diversity within or between cell populations is still heavily undercharacterised. Mass spectrometry-based single-cell proteomics (SCP) offers novel opportunities to study proteome variation on the single-cell level. Such variation can, however, be attributed to various sources, both biological and technical in nature. To fully unlock the quantitative potential of SCP, novel data normalization approaches must be implemented to separate the biological variation of interest from other confounding factors. This requires the assessment of such confounders, both biological and technical in nature, such as variation in digestion, matrix-associated ionization suppression and injection accuracy, cell size and morphology, and cell cycle.

To facilitate the development of different data normalization strategies and assess their performance, we propose the generation of empirical datasets containing several levels of ground truth data. This involves the use of DIGESTIF protein spike-ins to assess and correct for technical variation, as well as a Small Molecule-Assisted Shutoff (SMASh)-tag system to generate proteome-based readouts for cell size across a panel of morphologically distinct cell lines in vitro. In conjunction with a Fluorescent Ubiquitination-based Cell Cycle Indicator (FUCCI) system, this allows to uncouple cell size from cell cycle effects and to individually assess their contributions on proteome complexity and data normalization. Such reference data are expected to help develop new quantitative normalization models for SCP data, in particular towards heterogenous datasets comprising multiple cell types.