Description
Expose uncertainty, missing edits, convergent scars, and model assumptions as first-class outputs in lineage reconstruction pipelines. Boundary domains: machine-learning, phylogenetics. Representative papers: A Bayesian phylodynamic inference framework for single-cell CRISPR/Cas9 lineage tracing barcode data with dependent target sites.
Resolution criteria
Resolution requires: (1) Lineage reconstruction pipeline with uncertainty quantification for >=90% of tree edges; (2) Detection of >=95% of simulated convergent scars with FDR <=5%; (3) Benchmark dataset with ground truth for >=5 scenarios; (4) KG edges capturing reconstruction confidence. Deliverable: pipeline + benchmark dataset.