ABySS 2.0: resource-efficient assembly of large genomes using a Bloom filter

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• De novo DNA sequence assembly is crucial for genomics research and downstream applications like variation analysis.
• Rapid increases in DNA sequencing throughput necessitate efficient, scalable, and accurate bioinformatics tools for genome assembly.
• Previous methods like ABySS 1.0 required substantial memory, limiting accessibility.

Purpose of the Study:

• To redesign the ABySS software for more efficient and scalable de novo genome assembly.
• To reduce the memory footprint of genome assembly tools.
• To enable high-quality draft genome generation on widely accessible computing resources.

Main Methods:

• Implemented a de Bruijn graph representation using Bloom filters to decrease memory usage, departing from MPI.
• Benchmarked ABySS 2.0 using Illumina paired-end and mate-pair libraries from the Genome in a Bottle dataset.
• Investigated the integration of BioNano Genomics and 10x Genomics Chromium data to enhance assembly contiguity.

Main Results:

• ABySS 2.0 achieved a scaffold contiguity of 3.5 Mbp (NGA50: 3.0 Mbp) using less than 35 GB of RAM.
• The memory requirement is significantly reduced, often fitting within a single computer's capacity.
• Incorporating long-read technologies improved scaffold contiguity to 42 Mbp (NGA50: 15 Mbp).

Conclusions:

• ABySS 2.0 offers a memory-efficient and scalable solution for de novo genome assembly.
• The tool facilitates the generation of high-quality draft genomes, supporting large-scale genomics initiatives.
• Further improvements in assembly contiguity are achievable with the integration of long-read sequencing data.