Identifying and classifying trait linked polymorphisms in non-reference species by walking coloured de bruijn graphs

Area of Science:

• Genomics
• Bioinformatics
• Population Genetics

Background:

• Single Nucleotide Polymorphisms (SNPs) are crucial for understanding inheritable traits and gene identification.
• Current polymorphism detection methods often require a reference genome for sequence alignment, limiting their application in unsequenced organisms.
• Rapid generation of marker libraries is essential for timely gene discovery.

Purpose of the Study:

• To develop and present Bubbleparse, a novel method for detecting genetic variants directly from next-generation sequencing reads without a reference genome.
• To demonstrate the sensitivity, speed, and ease of use of the Bubbleparse algorithm.
• To evaluate Bubbleparse's performance against reference-based methods and showcase its utility in unsequenced species.

Main Methods:

• Utilized the de Bruijn graph implementation within the Cortex framework.
• Developed an algorithm to identify 'bubbles' in de Bruijn graphs, representing polymorphisms.
• Applied heuristics to optimize the detection of true polymorphisms.

Main Results:

• Bubbleparse demonstrated high sensitivity in detecting numerous polymorphisms quickly.
• Performance was comparable to reference-based polymorphism detection methods in Arabidopsis thaliana.
• Successfully applied Bubbleparse to unsequenced wild potato relatives, enabling rapid identification of disease resistance genes.

Conclusions:

• Bubbleparse offers a fast, sensitive, and reference-free approach for variant detection from next-generation sequencing data.
• The method is particularly valuable for studying organisms lacking a reference genome, such as wild crop relatives.
• Bubbleparse facilitates rapid genetic marker discovery and gene identification, accelerating research in plant genetics and breeding.