CINECA - Meta-analysis of splicing and expression QTLs across three cohorts D4.4

The aim of the CINECA project is to develop common infrastructure to support federated data analysis across national cohorts in Europe, Canada and Africa. In this deliverable, we report on the progress that we have made over the past four years to streamline the federated genetic analysis of molecular traits across multiple international cohorts.

The previous state of the art for federated genetic analysis of molecular traits is exemplified by the approach taken by the eQTLGen consortium, which relied on manual installation of software dependencies, manual execution of complex analysis steps, and sharing of large summary files between cohorts. Inspired by these limitations, we have designed and implemented six modular workflows to quantify and normalise molecular traits (such as gene expression or RNA splicing levels) from sequencing data (workflows 1-2), pre-process genotype data (workflow 3) and test for associations between molecular traits and genotypes (workflows 4-6). Our approach improves on the previous state of the art by packaging software dependencies into Docker/Singularity containers, using the Nextflow language to orchestrate complex multi-step workflows, and using the HASE(1) framework to significantly reduce the amount of data that need to be transferred between cohorts.

To encourage the adoption and re-use of our workflows by third parties, we have developed comprehensive training materials and prepared realistic open access datasets to test execution in diverse compute infrastructures. Finally, we demonstrate how our workflows can be used to perform federated analysis across multiple real cohorts located in Switzerland, Germany, the Netherlands and Estonia.