Skip to main content
Fig. 1 | BMC Genomics

Fig. 1

From: A graphical, interactive and GPU-enabled workflow to process long-read sequencing data

Fig. 1

Screenshots of our interactive GPU workflow which uses the Biodepot-workflow-builder platform. Panel A is a screenshot of the workflow using the open-source Bonito basecaller. Panel B is a screenshot of the workflow using the proprietary Guppy basecaller. Both basecallers use GPUs. For the Guppy workflow, the user enters the URL for the Oxford Nanopore Technology Guppy installation package which is then used to create a container to execute Guppy. The other steps in the two workflows are identical, consisting of data download, alignment and visualization. Each of these steps are performed by software modules encapsulated in Docker containers and represented by the graphical widgets. Lines connecting the widgets indicate flow of data between the modules. The user double clicks on the Start widget, enters the necessary parameters into the forms and presses a graphical start button to start the workflow. Double-clicking on a widget brings up a point-and-click interface for users to enter parameters, monitor results and control execution of the associated workflow module. Unlike other workflow execution platforms, the Biodepot-workflow-builder supports modules with interactive graphics. This is leveraged in this workflow to automatically open the final BAM files in the Interactive Graphics Viewer (IGV) which we use to check for diagnostic translocation breakpoints in our cell-line data. The execution time of the basecallers Guppy and Bonito on GPU-enabled machines using the NB4 cell line averaged 88.9 s (standard error 1.2) and 948.2 s (standard error 1.7) on an AWS g4dn.4xlarge GPU instance. For comparison, the CPU version of Guppy averaged 2551.8 s (standard error 22.4) on an AWS virtual machine instance (c5d.18xlarge) using 72 vCPUs

Back to article page