Accelerated diagnostics through hardware acceleration
Populärvetenskaplig sammanfattning av projektet
Whole genome sequencing (WGS), is the method of choice in the diagnostics of rare genetic diseases. WGS facilitates the detection of a diversity of variation in a single experiment, ranging from single nucleotide variants (SNV), to large copy number variation (CNV). In Addition, WGS allows for the unbiased search of disease-causing variation across the entire genome: which is of great importance in rare diseases, where the causative gene commonly is not known beforehand.
The cost of WGS has dropped dramatically over the years, and currently a patient genome may be sequenced and processed for roughly 25000 SEK, in less than five days’ time. Although the costs of WGS are at an all-time low, the costs are still prohibitive, limiting the number of patients that may be diagnosed through WGS.
The WGS analysis is performed on specialized high-performance computational clusters, and the standard computational pipelines completes in roughly 2 day’s time, producing over 100 Gb of data per patient. The computational costs are therefore greatly contributing to the overall costs of the entire WGS workflow. Hardware acceleration is the use of highly efficient hardware, such as graphical processing units (GPU) and field-programmable arrays (FPGA). These solutions are promising, as they allow for cheaper and more efficient analyses. In this project, we aim to develop and evaluate hardware acceleration in rare disease diagnostics; including software designed to be run on GPU, as well as on FPGA (including Dragen). The overall aim of the project is to optimize the current computational WGS workflows; improving the availability of the top genetic diagnostics.