Oct 31 , 15:30 - 16:30
nf-co2footprint: A Nextflow plugin to estimate the carbon footprint of pipeline runs
Data analysis using computational pipelines is scaling to new heights, allowing many fields to unlock new depths of interpretation. However, pipeline optimisations tend to focus on speed rather than resources, making CO2 emissions a neglected issue for many applications.
We aim to make Nextflow pipelines natively report on the CO2 footprint of the current workflow, enabling users to make informed decisions about the allocation of computational resources and to adopt more eco-friendly configurations. During pipeline execution, nf-co2footprint calculates the CO2 emissions for each task based on Nextflow's resource usage metrics and information about the power consumption of the underlying compute system, taking into account the carbon footprint of the respective energy production. All results are presented in a user-friendly report similar to the Nextflow pipeline execution report.
In a prototype version of the nf-co2footprint, we have already demonstrated the potential of carbon footprint analysis to further optimise resource usage. We will present the latest version of nf-co2footprint, among others with improved provenance tracking, and share updates about the project. We aim to improve its accuracy and make it more widely applicable to researchers and institutes around the world using different compute systems.
View project
Speaker
Co-authors
Sabrina Krakau, Till Englert, Júlia Mir-Pedrol, Phil Ewels and Sven Nahnsen