Michigan Imputation Server: Migrating an important resource for human genetics to Nextflow

The Michigan Imputation Server (MIS) is a web-based service for genotype imputation aimed at democratizing access to large reference panels. Since its start in 2015, the service has imputed over 112 million genomes for more than 12,000 users, been recognized as an important resource for human disease genetics with over 3,000 citations and has become the backbone for genome-wide association studies (GWAS). Behind the scenes, the server divides reference and target data sets into small chromosome segments that are processed in parallel. Originally developed for Hadoop MapReduce, this talk outlines how we successfully migrated the full imputation pipeline - including validation, quality control, phasing and imputation - to Nextflow, and highlights lessons learned. The publicly available Nextflow pipeline will enable others to setup their own imputation servers or run a proven imputation pipeline on different cluster architectures or cloud providers.