Genomic Variant Compression Using Phased Haplotype Mosaics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current data compression methods for genomic data, such as VCF files, are inefficient and require large storage and computational resources, especially in environments with limited resources like laptops and smartphones, due to high redundancy and computational expense of phasing information.
Innovation Solution
The method involves identifying and encoding two mosaics of segments from a reference database of phased haplotypes that match an individual's genome, incorporating phase information to improve matching accuracy and using a positional Burrows-Wheeler transform to efficiently optimize the compression process, balancing haplotype switches and mismatches to minimize the compressed representation size.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If standard compression tools (bgzip, M3VCF) are used on VCF files, then the file size is reduced to some extent, but the compression ratio is insufficient and files remain 100-500 Mb in size
Solution Approach 1:
The genome is divided into two separate haplotype sequences (maternal and paternal chromosomes), each processed independently through the compression algorithm. This segmentation allows the method to exploit haplotype-specific patterns and achieve better compression than treating the genome as a single undifferentiated sequence.
Solution Approach 2:
The method creates a compressed representation by copying segments from a reference genome and encoding only the deviations (mutations) from the reference. Instead of storing the entire genome sequence, it stores references to reference genome segments and a compact representation of differences, achieving high compression ratios.
2Productivity
If phasing information is incorporated to improve matching accuracy, then compression efficiency improves, but computational expense increases
Solution Approach 1:
Phasing information (haplotype assignment) is computed and stored in advance during the compression process. This preliminary action allows the decompression process to simply look up pre-computed haplotype matches rather than performing complex phasing calculations, shifting computational burden to the compression phase where it can be optimized and cached.
Solution Approach 2:
The method replaces complex real-time phasing calculations with a lookup-based system using pre-phased reference haplotypes. Instead of mechanically computing phase information during decompression, the system substitutes this with efficient pattern matching against stored reference sequences, reducing computational expense.
3Quantity of substance
If high compression is achieved to reduce storage requirements, then suitability for mobile devices improves, but indexability and random access capability deteriorate
Solution Approach 1:
The compressed format serves multiple functions: it provides high compression for storage efficiency, maintains indexability for random access, and enables both whole-genome and regional queries. The reference-based structure allows the same compressed representation to support multiple operational modes without sacrificing any single function.
Solution Approach 2:
The reference genome acts as an intermediary between the compressed data and the user. Instead of directly indexing the compressed haplotype differences, the system indexes the reference genome segments, allowing random access through the reference structure while the compressed differences provide the actual variant information.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
Methods and apparatus for compressing and decompressing genetic information from an individual are disclosed. In one arrangement, a data compression method generates a compressed representation of at least a portion of an individual's genome. The method comprises receiving an input file comprising a representation of the at least a portion of the individual's genome in the form of a sequence of variants defined relative to a reference genome. A reference database comprising a plurality of reference lists of genetic variants from other individuals is accessed. Each reference list comprises a sequence of genetic variants from a single, phased haplotype. Two mosaics of segments from the reference lists are identified which match the at least a portion of the individual's genome to within a threshold accuracy. Each mosaic represents a single one of the two haplotypes of the individual's genome in the at least a portion of the individual's genome for which the compressed representation is to be generated. Each of the segments comprises a portion of the sequence of genetic variants from one of the reference lists. The compressed representation is generated by encoding the two mosaics and deviations from the two mosaics.