G-ALN: a GPU-accelerated BWA-aln algorithm for palaeogenomics
Abstract Summary: The most commonly used method to achieve optimal mapping for short and damaged ancient DNA (aDNA) reads is the BWA-aln algorithm with seeding disabled. However, this CPU-based algorithm is computationally demanding, as the aligner must tolerate mismatches and indels across the entire read. To improve alignment speed, we present G-ALN: a GPU implementation of the BWA-aln and…
In the field of palaeogenomics, researchers have developed a new tool called G-ALN, which stands for GPU-accelerated BWA-aln algorithm for palaeogenomics. This innovative software aims to enhance the speed of aligning short and damaged ancient DNA (aDNA) reads, a process that is typically computationally demanding when using the BWA-aln algorithm on a CPU.
G-ALN takes advantage of the powerful parallel processing capabilities of GPUs to significantly speed up the alignment of aDNA reads. By first searching all reads on GPU cores, the algorithm identifies those that require a long search and dynamically shares these tasks between the GPU and any available CPU cores. This approach ensures that every read is searched using the same algorithm as the original BWA, resulting in byte-identical output.
The developers of G-ALN have also confirmed the accuracy of their algorithm using woolly mammoth and Neanderthal datasets, which are specifically relevant to aDNA studies. In benchmarking tests, G-ALN demonstrated a remarkable improvement in performance. The software indexed a reference genome approximately 32 times faster than the original BWA algorithm and increased the read alignment speed by about six times.
The development of G-ALN was facilitated by AI-assisted coding using Claude Opus 5.5, a powerful artificial intelligence tool. The resulting software is designed to process thousands of reads in parallel on each GPU, ultimately aligning more than twice as many aDNA reads per unit of energy compared to the CPU-based BWA-aln algorithm.
G-ALN is implemented in C++ and CUDA programming languages, making it accessible to researchers with the necessary expertise. The software is freely available under the GPLv3 license, ensuring that it can be used and adapted by the wider scientific community. It requires an NVIDIA GPU with compute capability 7.0 or higher and CUDA 11.8 or later to function optimally.
By offering a practical solution for leveraging GPU resources in established aDNA analysis workflows, G-ALN represents a significant advancement in the field of palaeogenomics. Researchers can now harness the power of GPUs to accelerate their analysis of ancient DNA, paving the way for more efficient and insightful studies of our evolutionary past.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
