Why accelerate genomics workflows?

  • In the last couple of years, the field of AI has shown the importance of GPUs in accelerating machine learning workflows

  • Current NGS analysis workflows can take days (especially for large datasets)

Danger

GPUs for NGS - Can we use GPUs to accelerate NGS workflows and reduce the

processing time to hours/minutes?

Note

CPUs vs GPUs

CPU

GPU

image1

image2

Core (processing units) can complete a variety of tasks one after another very quickly

Core (processing unit) specialized to do small, specific task

Several cores (e.g AMD EPYC 7642 48-Core)

Many cores (e.g., NviDIA H100 15,432)

Best for Serial processing tasks

Best for highly parallel tasks

Traditional programs are written for CPU processing

Additional Software are required to convert CPU functions to GPU functions for parallel execution