BiocNeighbors

BiocNeighbors implements exact and approximate methods for nearest neighbor detection in a framework that allows them to be easily switched within Bioconductor packages or workflows.

Key Features:

• Exact Search Methods: Exact searches can be performed using the k-means for k-nearest neighbors algorithm, vantage point trees, or an exhaustive search.
• Approximate Search Methods: Approximate searches can be performed using the Annoy or HNSW libraries.
• Configurable Search Parameters: Each search can be performed with a variety of different distance metrics, parallelization, and variable numbers of neighbors.
• Range-Based Searches: Range-based searches to find all neighbors within a certain distance are supported.

Scientific Applications:

• Bioconductor Integration: Enables nearest neighbor detection to be easily switched within Bioconductor packages or workflows.

Methodology:

Exact searches use the k-means for k-nearest neighbors algorithm, vantage point trees, or exhaustive search; approximate searches use the Annoy or HNSW libraries.