Genomic Read Compression Using CIGAR-Based Difference Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-throughput sequencing technologies generate vast amounts of genomic data, leading to significant IT costs due to storage and transmission requirements, with existing compression methods failing to effectively reduce redundancy in mapped and aligned sequencing data.
Innovation Solution
A method that compresses genomic data by unwinding nucleotide sequences using mapping positions and CIGAR strings, computing differences with previous reads, and passing these differences through an entropy coder for lossless compression, utilizing a short-time memory for local implicit assembly without requiring a reference genome, and allowing non-sequential access to compressed data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If general-purpose compression methods are used on genomic data, then compression is achieved, but compression efficiency is insufficient due to redundancy in mapped reads
Solution Approach 1:
The patent segments the genomic data processing into distinct components: unwinding the nucleotide sequence using CIGAR string operations, computing differences between unwound sequences, and applying entropy coding separately to these difference values. This segmentation allows each component to be optimized independently, achieving better compression than general-purpose methods that treat the data as a monolithic stream.
Solution Approach 2:
The patent transforms the representation of genomic data by changing parameters: converting mapped reads into unwound nucleotide sequences, then computing difference values between consecutive unwound sequences. This parameter transformation reveals the redundancy structure in mapped reads, enabling the entropy coder to achieve superior compression ratios compared to applying compression directly to the original formatted data.
2Reliability
If mapped and aligned reads are stored in detail, then complete alignment information is preserved, but storage size increases significantly
Solution Approach 1:
The patent extracts only the essential variation information from complete aligned reads by computing differences between consecutive unwound sequences. Instead of storing full alignment details for each read, the method stores difference values that can be reconstructed into the original data through reverse computation, significantly reducing storage requirements while preserving complete information.
Solution Approach 2:
The patent inverts the traditional storage approach by not storing the reads directly but storing the differences between reads. This inversion allows the system to reconstruct the original reads on-demand while using minimal storage space, as the difference values are much more compact than the full read sequences.
3Quantity of substance
If compression methods use reference genomes, then compression accuracy improves, but processing complexity and time increase
Solution Approach 1:
The patent implements a self-service compression approach where the algorithm uses only the input reads themselves to compute differences, without requiring external reference genomes or additional biological knowledge. The compression is achieved by exploiting the natural redundancy between consecutive reads in the input data, making the system self-contained and reducing processing complexity.
Data Source
AI summary
The present invention relates to a method for compressing genomic data, whereby the genomic data are stored in at least one data file containing at least a plurality of reads built by a genome sequencing method, whereby each read includes a mapping position, a CIGAR string and an actual sequenced nucleotide sequence as a local part of the donor genome, comprising the steps: —unwind a nucleotide sequence of a current read of one of said data files by using the mapping position and the CIGAR string of said current read, whereby said current read has at least one previous read, —compute a difference between the unwound nucleotide sequence of said current read and an unwound nucleotide sequence of at least one of said previous reads, whereby said difference contains the differences of the mapping positions and the nucleotide sequences, —pass said computed difference to an entropy coder to compress said difference, —encode said current read by the compressed difference, and —repeat the forgoing steps with said current read as one of said previous reads and a following read as a new current read until no more following reads are available.