Single-Cell Genomics Subsampling via Prize Collecting Steiner Tree
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for downsampling large-scale single-cell genomics datasets are inefficient in terms of memory usage and often result in suboptimal representation of the dataset manifold, leading to unnecessary data processing time and loss of information.
Innovation Solution
A computer-implemented method that generates a cell-cell neighborhood graph, divides cells into seed and non-seed cells, and uses a prize collecting Steiner tree algorithm to select a subsample of cells, preserving the original dataset manifold and reducing memory requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If random sampling is used to reduce dataset size, then processing time is reduced, but the representation of the dataset manifold becomes suboptimal and information is lost
Solution Approach 1:
The patent segments the single-cell genomics dataset into multiple partitions that can be processed independently. Each partition is handled separately through memory mapping, allowing the system to work with manageable subsets of data while preserving the overall dataset structure and information content.
Solution Approach 2:
The patent introduces memory-mapped files as an intermediary between the raw dataset and the processing algorithms. This intermediary layer allows efficient access to large datasets without loading them entirely into memory, enabling both fast processing and complete information retention through lazy loading of data blocks.
2Measurement precision
If large-scale computing infrastructure is used to process complete datasets, then analysis accuracy is improved, but computational resource requirements increase
Solution Approach 1:
The patent divides the large-scale dataset into smaller partitions that can be processed on standard computing devices. This segmentation allows accurate analysis to be performed on manageable data subsets without requiring complex large-scale computing infrastructure.
Solution Approach 2:
The patent processes data in partial increments through memory-mapped partitions, loading only the necessary portions into memory during processing. This approach achieves accurate analysis results without requiring the entire dataset to be loaded, reducing computing infrastructure requirements while maintaining analysis precision.
3Loss of information
If complete datasets are processed to maintain information content, then information retention is improved, but memory requirements increase
Solution Approach 1:
The patent uses memory-mapped files as an intermediary that provides virtual access to the complete dataset without requiring physical loading into memory. The operating system's virtual memory management handles data paging between disk and memory, allowing complete information retention with minimal actual memory usage.
Solution Approach 2:
The dataset is segmented into memory-mapped partitions that are loaded into memory only when needed. This segmentation allows the system to retain access to all data information while using minimal memory at any given time, as only active partitions occupy physical memory resources.
Data Source
AI summary
The present disclosure relates to a computer-implemented method of extracting a subsample of cells from a plurality of cells in a sin-gle-cell genomics dataset, the method comprising the steps of: obtaining a single-cell genomics dataset represented in at least two dimensions, wherein information about each cell is represented in a first dimension and information about genomic features is represented in a second dimension; generating a cell-cell neighborhood graph from the single-cell genomics dataset, the cell-cell neighborhood graph providing information about similarities of the genomic features of the cells, wherein the cells are represented as vertices in the cell-cell neighborhood graph; dividing the cells in the cell-cell neighborhood graph into seed cells and non-seed cells; assigning at least one first prize to the seed cells and at least one second prize to the non-seed cells in the cell-cell neighborhood graph; and traversing the cell-cell neighborhood graph using a prize collecting steiner tree algorithm to obtain a subsample of cells. The present disclosure further relates to a computer system for extracting a subsample of cells from a plurality of cells in a sin-gle-cell genomics dataset.