LAML-Pro: Joint Maximum Likelihood Inference of Cell Genotypes and Cell Lineage Trees

Area of Science:

• Computational Biology
• Genomics
• Cell Biology

Background:

• Dynamic lineage tracing uses genome editing to create heritable mutations for reconstructing cell lineages.
• Current methods separate genotype identification from lineage tree inference, leading to inaccuracies due to genotyping errors (≈25-50% in imaging).

Purpose of the Study:

• To develop a computational method that jointly infers cell genotypes and lineage trees, improving accuracy in cell lineage reconstruction.
• To address the challenge of genotype errors in dynamic lineage tracing data.

Main Methods:

• Introduced LAML-Pro (Lineage Analysis via Maximum Likelihood with PRobabilistic Observations), a novel algorithm based on the Probabilistic Mixed-type Missing Observation (PMMO) model.
• The PMMO model accounts for both genome editing and genotype observation processes.
• Leveraged model sparsity for efficient lineage tree construction from thousands of cells.

Main Results:

• LAML-Pro corrects genotype errors and infers more accurate cell lineage trees compared to existing methods.
• Reduced genotype errors by 5-fold in imaging-based lineage tracing data.
• Produced more spatially coherent lineage trees.

Conclusions:

• LAML-Pro offers a significant advancement in cell lineage tracing by integrating genotype inference and tree reconstruction.
• The method improves the accuracy and reliability of lineage trees derived from dynamic lineage tracing experiments.
• LAML-Pro is available for broader research use.