bgnorm: A Generative Statistical Framework for Background Correction, Normalisation, and Quality Control in Multiplex Spatial Proteomics

Authors

Kharbanda, M., Tubelleza, R., Tan, Y., Tan, C. W., Janke, C., Sebina, I., Belz, G., Kulasinghe, A., Salim, A., Bhuva, D. D.

Abstract

Multiplex spatial proteomics enables highly multiplexed in situ profiling but remains limited by technical variation arising from autofluorescence, non-specific antibody binding, instrument noise, and staining variability, affecting downstream biological tasks like cell typing. We present bgnorm, a statistical framework that describes fluorescence measurements using a generative mixture model of background, non-specific binding, and biological signal components. As natural statistical consequences, the model yielded three new methods: a background-correction method through probabilistic deconvolution of protein intensities, quality control metrics, and a quantile normalisation approach to unify measurements across markers, samples, and sequential slices. Across multiple multiplex imaging technologies, bgnorm improves signal separation and downstream marker positivity classification compared with existing preprocessing approaches. In expert-annotated datasets comprising over 406,000 marker positivity annotations, bgnorm achieved the highest classification performance and enabled accurate use of a single global positivity threshold across markers and samples. The method is implemented in the bgnormR and bgnormpy packages.

Preprint server: bioRxiv
The authors list and abstract were imported from bioRxiv on 12 Aug 2026.

Stats

1-terrible, 9-excellent. How would you rate this preprint? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.