Partition

Partition performs Bayesian model-based clustering of individuals from out-crossing diploid organisms using genotypes at co-dominant marker loci to identify population subdivisions and assign individuals probabilistically.

Key Features:

• Population Genetic Model: Employs an underlying genetic model suitable for analyzing out-crossing diploid species.
• Clustering Capabilities: Clusters individuals into outcrossing populations, hybrid generations, full-sib families, and selfing lines based on genotypes at co-dominant marker loci.
• Bayesian Approach: Utilizes a fully Bayesian framework that represents uncertainty about sample partitions as probability distributions over possible configurations.
• Exact Linkage Algorithm: Applies an agglomerative hierarchical exact linkage algorithm, a specialized form of the maximin clustering algorithm, to transform posterior co-assignment probabilities for visualization.
• Visualization through PartitionView: Produces a rooted binary tree or a forest in PartitionView in which each node represents a set of individuals and node height indicates the posterior co-assignment probability.

Scientific Applications:

• Population structure inference: Identification of sub-divisions and patterns of genetic variation in population genetics and evolutionary biology.
• Hybridization analysis: Detection and characterization of hybrid generations.
• Kinship and family structure: Assignment and identification of full-sib families within populations.
• Selfing lineage analysis: Examination of dynamics and structure of self-fertilizing lineages.

Methodology:

Implements a fully Bayesian statistical framework representing partitions as probability distributions and applies an agglomerative hierarchical exact linkage (maximin-derived) algorithm to convert posterior co-assignment probabilities into rooted binary trees or forests for visualization in PartitionView.