Efficient Reconstruction of Stochastic Pedigrees: Some Steps From Theory to Practice

Area of Science:

• Population Genetics
• Computational Biology
• Genomic Inference

Background:

• Assessing ancestral information from extant populations is crucial for understanding evolutionary history.
• Previous work introduced REC-GEN, an algorithm for pedigree reconstruction under simplifying assumptions.
• The performance of REC-GEN on realistic, simulated data with inbreeding remains largely unexplored.

Purpose of the Study:

• To evaluate the REC-GEN algorithm's effectiveness on simulated population data.
• To identify limitations of REC-GEN, particularly concerning inbreeding.
• To develop an improved algorithm for accurate pedigree reconstruction in the presence of inbreeding.

Main Methods:

• Simulation of population data under various genetic models.
• Optimization of the REC-GEN algorithm's runtime.
• Comparative analysis of REC-GEN and a novel Belief-Propagation heuristic.
• Assessment of ancestral reconstruction accuracy in idealized versus inbred populations.

Main Results:

• The optimized REC-GEN algorithm showed poor performance in reconstructing pedigrees beyond two generations due to inbreeding.
• Ancestral reconstruction, a key component of REC-GEN, performed well without inbreeding but poorly in random mating populations.
• The novel Belief-Propagation heuristic demonstrated significantly improved performance in simulations incorporating inbreeding.

Conclusions:

• Inbreeding poses a significant challenge to existing pedigree reconstruction algorithms like REC-GEN.
• A Belief-Propagation approach effectively addresses inbreeding, leading to more accurate pedigree inference.
• This work provides a more robust method for reconstructing complex ancestral histories from genetic data.