Bayesian inference of lineage trees by joint analysis of single-cell multimodal lineage-tracing data with BiLinT

Abstract

The advent of single-cell lineage-tracing technologies has enabled simultaneous measurement of gene expression and lineage barcodes. However, integrating these modalities for high-resolution lineage reconstruction has remained challenging due to limitations of methods that analyze modalities separately. In response, we present BiLinT, a Bayesian framework for reconstructing high-resolution cell lineage trees by jointly modeling single-cell multimodal lineage-tracing data. BiLinT integrates lineage barcode evolution (modeled via a continuous-time Markov chain) and gene expression dynamics (modeled via an Ornstein-Uhlenbeck process) within a unified probabilistic model. Applications to synthetic and real datasets demonstrate improved accuracy and reveal developmental fate biases.