Hopper

Hopper performs iterative sampling to enable fast, targeted multi-resolution analysis of diverse cell types in large-scale single-cell RNA-seq (scRNA-seq) datasets.

Key Features:

• Iterative Sampling: Adds points iteratively during the sampling process to build representative subsets of cells.
• Multi-resolution Analysis: Enables fast and targeted multi-resolution analyses of diverse cell types within very large datasets.

Scientific Applications:

• Identifying Small Cell Populations: Detects small but biologically significant populations, exemplified by recovering a cluster of 64 macrophages expressing inflammatory genes in a reduced sample of 5,000 cells from over 1.3 million mouse brain cells.

Methodology:

Hopper uses the greedy k-centers algorithm with iterative farthest-first traversal to ensure a 2-approximation to the optimal Hausdorff distance between the full dataset and its downsampled version, preserving transcriptional diversity effectively.