Genomic DAG Serialization for Cross-System Portability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing approaches to structuring genomic sequences in databases restrict flexibility, as they often rely on hardware-specific implementations that cannot be easily copied or moved between systems, particularly due to the use of pointers and endianness, which are location-specific and not serializable.
Innovation Solution
The method involves serializing graph data structures representing biological information into a stream of bytes that can be streamed to a new location, allowing for deserialization into a clone with the same functionality and representation, enabling flexible transfer and use across different systems and storage mediums.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If graph data structures with pointers are used to represent biological information in memory, then access speed and traversal efficiency are improved, but portability and transferability between systems are worsened due to location-specific pointer values
Solution Approach 1:
The patent creates a serializable copy of the graph data structure by replacing location-specific pointers with relative offsets or indices that can be reconstructed in any memory layout. The serialization process captures the graph's topology and data without embedding system-specific memory addresses, allowing the structure to be copied and transferred between different computer systems while maintaining its functional integrity
Solution Approach 2:
The patent transforms the representation parameters of graph data structures by converting absolute memory addresses into relative offsets or index-based references. This parameter transformation allows the same logical structure to be instantiated with different physical memory layouts on different systems, thereby improving portability without sacrificing access efficiency
2Adaptability or versatility
If complex graph structures with interconnected pointers are implemented, then data representation flexibility and query capability are improved, but difficulty of copying and serialization are worsened due to interdependent pointer relationships
Solution Approach 1:
The patent segments the serialization process into distinct phases: first capturing the graph's topology structure, then recording data payloads, and finally establishing the mapping between them. This segmentation allows complex interconnected structures to be serialized independently in manageable components, reducing the overall complexity of the serialization operation while preserving the full representational flexibility of the original graph
3Speed
If native pointers are used for graph node references, then memory access efficiency is improved, but compatibility across different endianness systems is worsened
Solution Approach 1:
The patent introduces an intermediary serialization format that acts as a buffer between the native pointer-based graph structure and the target system's memory layout. This intermediary format uses platform-independent representations (such as standardized integer indices or relative offsets) that can be correctly interpreted regardless of the destination system's endianness, thereby maintaining both access efficiency and cross-platform compatibility
Data Source
AI summary
Methods of the invention include representing biological data in a memory subsystem within a computer system with a data structure that is particular to a location in the memory subsystem and serializing the data structure into a stream of bytes that can be deserialized into a clone of the data structure. In a preferred genomic embodiment, the biological data comprises genomic sequences and the data structure comprises a genomic directed acyclic graph (DAG) in which objects have adjacency lists of pointers that indicate the location of any object adjacent to that object. After serialization and deserialization, the clone genomic DAG has the same structure as the original to represent the same sequences and relationships among them as the original.