When computational pipelines go 'clank'

Vivien Marx
2020 Nature Methods  
love the smell of fresh baked Snakefiles in the morning!" tweeted University of California, Davis, computational biologist C. Titus Brown earlier this year. Brown and his team, including postdoctoral fellow N. Tessa Pierce, are putting together a computational pipeline for decontaminating genome assemblies in metagenomics and are using the workflow manager Snakemake 1 to do so. It's one of over 100 workflow manager tools pipeline builders might use. Without data analysis pipelines, many 'big
more » ... a' fields would move dramatically slower, says UCLA cancer researcher and computational biologist Paul Boutros. Pipelines and their infrastructure become joint workspaces for computational biologists and software engineers, he says. Not all pipelines are as standardized as, say, the Genome Analysis Tool Kit steps for mapping sequence to a reference. The choice of pipeline can lead to differing results on the same data, as a study in neuroimaging has revealed 2 . Using, building and maintaining pipelines presents researchers with a forked road. (You can read more on our blog.)
doi:10.1038/s41592-020-0886-9 pmid:32601427 fatcat:2w5qi2spgnasdl6eczgiwfft3a