Accurate and efficient cell lineage tree inference from noisy single cell data: the maximum likelihood perfect...

Area of Science:

• Computational biology
• Genomics
• Evolutionary biology

Background:

• Cell lineage trees trace cellular evolutionary history.
• Inferring these trees from noisy single-cell genotype data is computationally challenging.
• Existing methods often assume uniform genotype uncertainty and are slow.

Purpose of the Study:

• To develop a novel method, ScisTree, for inferring cell lineage trees and genotypes from noisy single-cell data.
• To address the limitation of non-uniform genotype uncertainty in real-world data.
• To provide a computationally efficient alternative to existing methods.

Main Methods:

• ScisTree utilizes genotype probabilities for individual genotypes, accommodating non-uniform uncertainty.
• It implements a fast heuristic under the infinite sites model.
• The method infers cell lineage trees and calls genotypes that maximize likelihood and allow perfect phylogeny.

Main Results:

• ScisTree demonstrates high accuracy in inferring cell lineage trees compared to existing methods.
• The method is significantly more efficient, enabling analysis of larger datasets.
• Its efficiency allows for new applications like doublet imputation.

Conclusions:

• ScisTree offers an accurate and efficient solution for cell lineage tree inference from noisy single-cell genotype data.
• The handling of individualized genotype probabilities is a key advancement.
• The tool's speed and accuracy open new avenues for single-cell data analysis.