Abstract
Despite recent advances in inferring cellular dynamics using single-cell RNA-seq data, existing trajectory inference (TI) methods face difficulty in accurately reconstructing cell-state manifold and inferring trajectory and cell fate plasticity for complex topologies. We present MARGARET, a novel TI method that utilizes a deep unsupervised metric learning-based approach for inferring the cellular embeddings and employs a novel measure of connectivity between cell clusters and a graph-partitioning approach to reconstruct complex trajectory topologies. MARGARET utilizes the inferred trajectory for determining terminal states and inferring cell-fate plasticity using a scalable absorbing Markov Chain model. On a diverse simulated benchmark, MARGARET out-performed state-of-the-art methods in recovering global topology and cell pseudotime ordering. When applied to experimental datasets from hematopoiesis, embryogenesis, and colon differentiation, MARGARET reconstructed major lineages and associated gene expression trends, better characterized key branching events and transitional cell types, and identified novel cell types, and branching events that were previously uncharacterized.
Competing Interest Statement
The authors have declared no competing interest.