yannick.wurm at unil.ch
Thu Jan 20 04:08:55 UTC 2011
Hello & thanks for the comments.
> Swappable/configurable steps might be
> Pre-process (trim / quality filters?)
> Alignment (align)
> Post alignment (gblocks)
> Translation (to Nexus)
> Analysis (Phyml)
> The idea is that we could swap in components (TNT or RaXML for Phyml,
> Muscle for MAFFT etc.)- but also that the pipeline remains "simple".
Yes, thats what I would ideally want. (as well as being able to easily modify the run options of the programs). How would you go about generalizing this?
Right now I'm basing "what do to" on the file extensions I provide... which limits me based on the file extensions...
> I think it's a great idea to generate predefined pipelines for common
> bioinformatics tasks. I experimented with a tool called Boson six months ago.
> It could be worth looking if you feel like investing more time into your
> Boson commands, similar to rake tasks, are more modular and can be installed
> from the web into a ~/.boson directory. This has obvious advantages over
> a single rake file. Boson tasks can be chained together where the data is
> passed around in YAML format.
> The github link is - https://github.com/cldwalker/boson
I haven't looked thoroughly now, but at least superficially, Boson looks real cool.
However, I'm a bit scared of investing energy into technologies that are too new. Boson has only one developer who may or may not keep his project alive over the next years. Time I invest in learning something today ... must continue to help improve my productivity over the next 5 or 10 years by still being reusable in 5 or 10 years (with as few modifications as possible). There is uncertainty to everything, but something like Boson does seems a bit too risky right now...
Ant Genomes & Evolution
More information about the BioRuby