Forum:Scripts on the Web: Bioinformatics Recipes

Question

[Istvan Albert]

This is an announcement of a new data analysis platform that we are developing.

We call the concept Bioinformatics Recipes. A demonstration version of the site is available at:

• https://www.bioinformatics.recipes/

In a nutshell, it is a platform for deploying and sharing scripts on the web. We call these scripts recipes. See the:

• Bioinformatics Cookbook

The recipes may be written in bash, R, makefiles, or via any other workflow engine. In addition the software supports data management both for the input data and for the results generated when executing the recipes.

We envision labs and individuals running their own sites to maintain the custom functionality that they need. The project is supported by the US Fish and Wildlife Service and will be designed to support sequencing facilities needing to manage data and data analysis pipelines.

---

The code that runs the site is published via:

• https://github.com/biostars/biostar-engine

the data analysis recipes are stored separately in:

• https://github.com/biostars/biostar-recipes

---

Additional links to docs:

• What are recipes?
• Concepts and terminology

[Devon Ryan]

Can you comment on how this differs from Galaxy and its workflows?

[Istvan Albert]

A few key differences:

1. It is not a workflow engine - workflows would be used within a recipe if someone whishes to do so.
2. A recipe is independent of the site. You can take a recipe and after minimal modifications run it from command line if you need to.
3. The recipes are scripts not wrappers around tools. A typical recipe involves many tools to produce a certain outcome.
4. The site is designed for users to take ownership, copy and modify existing recipes - make their own versions of it. Then they can reshare these. There could be dozens of variants for short read alignments even with the same tool. Or you could have short read aligner where the tool is selected as a parameter. (see the Short Read Alignment example in the Cookbook)
5. Users may choose to modify the interface for any recipe as they wish. The moment they copy a recipe it becomes their variant. The interface may be modified right away on the web.

Fundamentally the difference is in the philosophy of the approach.

Instead of championing a "standardized/centralized" approach - the recipe view recognizes, accepts and embraces that biology is complex and that we need to customize analyses.

The driving principle is that "we need to know what someone else did for the entire process" rather than having to piece that back together from individual tools. I ran tool A then ran tool B, then ran tool C.

A recipe is one, full, complete task from A to Z.