As recently demonstrated by the COVID-19 pandemic, large-scale pathogen genomic data are crucial to characterize transmission patterns of human infectious diseases. Yet, current methods to process raw sequence data into analysis-ready variants remain slow to scale, hampering rapid surveillance efforts and epidemiological investigations for disease control. Here, we introduce an accelerated, scalable, reproducible, and cost-effective framework for pathogen genomic variant identification and present an evaluation of its performance and accuracy across benchmark datasets of Plasmodium falciparum malaria genomes. We demonstrate superior performance of the GPU framework relative to standard pipelines with mean execution time and computational costs reduced by 27× and 4.6×, respectively, while delivering 99.9% accuracy at enhanced reproducibility.

Many traits of cancer progression (e.g., development of metastases or resistance to therapy) are facilitated by tumour evolution: Darwinian selection of subclones with distinct genotypes or phenotypes that enable such progression. Characterising these subclones provide an opportunity to develop drugs to better target their specific properties but requires the accurate identification of somatic mutations shared across multiple spatiotemporal tumours from the same patient. Current best practices for calling somatic mutations are optimised for single samples, and risk being too conservative to identify shared mutations with low prevalence in some samples. We reasoned that datasets from multiple matched tumours can be used for mutual validation and thus propose an adapted two-stage approach: (1) low-stringency mutation calling to identify mutations shared across samples irrespective of the weight of evidence in a single sample; (2) high-stringency mutation calling to further characterise mutations present in a single sample. We applied our approach to three-independent cohorts of paired primary and recurrent glioblastoma tumours, two of which have previously been analysed using existing approaches, and found that it significantly increased the amount of biologically relevant shared somatic mutations identified. We also found that duplicate removal was detrimental when identifying shared somatic mutations. Our approach is also applicable when multiple datasets e.g. DNA and RNA are available for the same tumour.