Evaluation of serverless computing for scalable execution of a joint variant calling workflow.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Whole-genome sequencing costs have decreased, but computational analysis remains a bottleneck.
• Serverless computing presents a potential alternative to traditional compute resources for genomic data analysis.

Purpose of the Study:

• To evaluate the utility and performance of serverless computing for standardized genomic workflows.
• To define and execute a best-practice joint variant calling pipeline using the SWEEP workflow management system.

Main Methods:

• Implemented the GATK best-practice short germline joint variant calling pipeline as a SWEEP workflow with 18 tasks.
• Executed the workflow on Illumina paired-end read samples from the 1000 Genomes project (European and African super populations).

Main Results:

• Workflow execution time ranged from approximately 3 hours (2 samples) to 13 hours (62 samples).
• Costs varied from $2 to $70, increasing linearly with sample size.
• Runtime was dominated by a single task for larger datasets.

Conclusions:

• Serverless computing is a viable and scalable option for executing standardized genomic workflows like joint variant calling.
• The SWEEP system effectively manages serverless execution of complex bioinformatics pipelines.