Anduril

Anduril 1.2.9 released

javi — Mon, 05 Mar 2012 08:54:27 +0000

The latest Anduril release 1.2.9 supports now multiline strings, several core bugs were also fixed.

Anduril bundle refactoring

Riku — Fri, 16 Dec 2011 11:03:30 +0000

Anduril components are currently structured in component bundles. While this has proven to be a convenient way so far, the expanding number of components and new bundles have made the component repository quite hard to handle – especially for if you’re new to Anduril! Moreover, with Anduril expanding to areas outside of the microarray research, adding new bundles has not been as efficient and scalable as we would like.

Until now Anduril releases have contained two bundles by default: builtin and microarray. Add-on bundles also include imaging, FACS and Moksiskaan bundles. Microarray bundle contains several components that are not specific to biological analysis and are also essential for any data analysis done with Anduril.

After some deliberation, we have decided to revamp our component structure in the coming months. We will retain bundles as deployment entities, but move from a bundle structure to a package structure. According to our current plan, the new packages will encompass

builtin (same as old builtin bundle)
util (general file conversion, csv handling, etc. components from the old microarray bundle)
- stat (Statistical testing)
- plot
- …
bio (e.g., most of old microarray specific components)
- Moksiskaan
- ACGH
- …
doc
- latex
- HTML
- plot
- …
imaging and any other add-on packages.

There is a lot of refinement to do. For example, the hierarchical structure will be expanded to include additional subpackages where we see fit. New packages can be added on any level of the hierarchy (e.g., Moksiskaan under bio, imaging on root level). Components will contain package declarations in their XML-interfaces. We are also planning to remove package (bundle) declarations out of the anduril run command line options into import statements within scripts. Import statements will be python/Java style.

Another big improvent will be the inclusion of datatypes into the package hierarchy. Currently data types are ‘flat’ and cannot overlap between bundles making inclusion of new data types in many bundles problematic. Once datatypes are tied to the hierarchy (e.g., ‘util.stat.CSV’), they can be uniquely identified.

During the transition phase, old style component calls can still be made e.g., myComp = LatexCombiner(param1=something). Component calls will be changed to follow something resembling the python standard, where importing is done by import doc.latex as latex and then component calls become myComp = latex.Combiner(param1=something). We will also provide a liftOver-script that will map legacy component calls in pipelines to new style calls.

Anduril @ Wikipedia

Riku — Tue, 15 Nov 2011 12:55:58 +0000

Prior to the release of the next version of Anduril, we also decided to (finally!) tell the greater non-scientific world about Anduril. How did we go about this? Check out Anduril on Wikipedia.

New Anduril 1.2.4 released with enhanced array support

javi — Fri, 16 Sep 2011 10:16:50 +0000

The new Anduril 1.2.4 is now released and includes array support for Matlab, Python and R components. See the release ChangeLog for further details.

Websites work again

Riku — Wed, 29 Jun 2011 06:02:36 +0000

All our websites and services (CANGES excluded) are up and running once more. It tooks us less time than we expected which is by no means a bad thing. Some performance issues and dead links persist but we are taking …

Web service downtime

Riku — Mon, 27 Jun 2011 12:37:28 +0000

Our old web server decided to terminate its contract during the weekend. Although this site did not go down, all the Anduril related websites did. We are rebuilding our services and anticipate to have everything up and running within the …

Anduril 1.2.1 with Remote Execution and Array Data Type released

javi — Thu, 16 Jun 2011 10:44:52 +0000

This new release (16 June 2011) allows for remote execution of components; the Anduril engine, installed in your own machine, manages via ssh the remote executions in pre-configured machines and can also interface with batch job schedulers for instance for cluster nodes. We also provide a new array data type which allows for a variable number of arguments in functions and components.

New Anduril 1.2.0 with dynamic for loops

javi — Mon, 16 May 2011 11:36:24 +0000

Anduril 1.2.0 is now available. Novel in this release are dynamic for loops in which we can iterate over an input that has been generated at run-time. We have also introduced a bind annotation that allows for configuring workflow dependencies …

TCGA commentary of Anduril

Riku — Fri, 06 May 2011 09:59:33 +0000

The Cancer Genome Atlas has an active public relations policy according to which they try to keep track and promote research articles that use TCGA data. Well, our Anduril article last year used the complete Glioblastoma data set from TCGA and they finally picked it up for their Research Highlights blog. The commentary makes a nice read on how to explain Anduril and results obtained using it to a layman .

Anduril 1.1.5 released

Riku — Fri, 08 Apr 2011 13:04:37 +0000

We’re pleased to end our week with a new release of Anduril. Anduril 1.1.5 contains several cool new features such as a for-loop, which is not a feature that workflow engines normally have. Ville also tweaked our microarray component API, which now contains a keyword search to help the user to find specific components. The search not only looks for component names but also parses the component descriptions. I personally feel the old “worse than Java”-API had it coming, and the new version is a terrific upgrade.

You can read about additional tweaks and fixes in the Anduril ChangeLog that is on the Download site.