For the upcoming Bio7 1.7 release i worked on a new general shell implementation. And with this implementation some new ideas in conjunction with R came to my mind.
So recently i integrated some new actions to ease the creation of R packages with R.

Please note: For the built process of R package binaries some tools are needed (especially for Windows). Please consult the R instructions on the internet or search for available tutorials.
For tests on the Windows OS i installed a Latex environment (MiKeTeX with inconsolata.sty style file installed) and the Rtools for Windows.

With the upcoming Bio7 release the following actions will be available:

In the file Navigator of Bio7 you can create a package skeleton from select R files or R workspace variables. By invoking the Create R Package Skeleton action a dialog pops up which offers options to create a package skeleton from the selected files or from availabe workspace variables and to enter a name for the package (see the next two screenshots below).

After the confirmation with “OK” the R package skeleton can be found as a subfolder of the selected Bio7 project. If you now select the created R skeleton folder and open the context menu of the file Navigator a new submenu is available with actions to check and built the R package (see screenshot below)

For the built process the new available native shell hast to be selected from the console menu of the Console view of Bio7 1.7.

If everything is configured properly (see the shell messages after the check or built process) a new built R package should be available in the R package folder (after a refresh of the folder).

If you would like to use the latest R version 3.0.0 with Bio7 on Windows you have to install Rserve in an R 3.0.0 installation and adjust some Bio7 preferences to use this installation instead of the bundled R binary.

Install Rserve in R 3.0.0 e.g. with the command

> install.packages(“Rserve”)

Alternatively you can use the RGUI menu to search and install Rserve.
Then you have to tell Bio7 the path to the install location of R 3.0.0:
Menu Preferences->Preferences Bio7->Path To R (see example Screenshot below).

Finally you have to adjust the path of the package location the path were additional downloaded R packages will be installed (see Screenshot below) :
Menu Preferences->Preferences Bio7->Preferences RServe->Package install location
Type in the R console of your R 3.0.0 installation e.g.:

> .libPaths()

to get the current path for the Bio7 “Package install location”!

If all paths are correct you can use the new R 3.0.0 installation from within Bio7.

I finished a rework of the image selection transfer from ImageJ to R. I splitted the action into three different actions for a better usability.
A new feature of Bio7 1.7 is the possibility to transfer multiple pixel selections (with a fixed or an increasing signature) from selected slices of stacks.
The transferred data then can e.g. be used for a supervised classification method in R were you can add image data selections with a fixed signature (e.g. representing a plant species) to already transferred selections (e.g. increasing signatures – different plants species) to improve the classification results.
Thanks to the Roi Manager of ImageJ selections can easily be saved and reloaded for a nice workflow in conjunction with the transfer of  the data to R (simply drag and drop ROI files on the ImageJ-Canvas to load or add them).

In addition i had also to rework the Image-Methods view for a better grouping of the different actions (see temporary labels in the screenshot below) which hopefully gives a better usability.

These are the last features (beside many others) for the next release and i hope that i can release the new version in a few weeks after some tests. If you have found some bugs in Bio7 1.6 don’t hesitate to contact me or write to the Bio7 forum.

See:
http://javarevisited.blogspot.de/2011/09/javalangoutofmemoryerror-permgen-space.html
http://javarevisited.blogspot.de/2011/05/java-heap-space-memory-size-jvm.html

I started to implement Jython support for the upcoming Bio7 1.7 release. With Jython it will be possible to use the Bio7 API’s and graphical interfaces with Python syntax. Furthermore it will be possible to execute Jython scripts in the Flow editor.

I changed the Bio7 website design for a better readability with mobile devices. I also embedded the Nabble Forum into the website structure and changed the default forum fonts.

“An Introduction to Scientic Image Data Analysis using R”.

Abstract:

The program of the ImageJ conference 2012 can be found here

In the next Bio7 release it will be possible to create custom GUI’s with the JavaFX Builder or with the new JavaFX API. I already integrated an API to load and compile a *.fxml file into a custom view of Bio7. GUI interfaces can easily be created with the new available JavaFX Scene Builder from Oracle and saved  in a project of the Bio7 workspace. Objects and actions are accessible by means of Dependency Injection using the @FXML annotation.

Furthermore in the next release several custom views can be created side by side (i removed the tab implementation) e.g. to combine several custom GUI’s and panels. In the screenshot below an example shows the created GUI of the Scene Builder (screenshot above)  side by side with three other custom views (processing API examples).

With Bio7 1.6 it is possible to send multiple images from ImageJ to R without the need to open them in the Graphical User Interface of ImageJ for speed improvements. With a simple script written in Java, Groovy or BeanShell a new Bio7 API command can be used (see below) to transfer images and e.g. to cluster image data with the available packages in R.

ImageMethods.imageToR(“currentImage”, true, 3, imp);  //The first argument is the transfer name of the image, the second argument creates a matrix if true (for datatype double and RGB transfer), the third argument determines the datatype for the transfer (byte, integer, double or RGB), the last argument expects an ImageJ image (can be opened with the ImageJ API) or if null transfers the current opened image in the Bio7 ImageJ panel (API).

The R commands itself can be embedded easily with the available Bio7 API or called from an external R script with the API. Here is a short video which uses the API to cluster selected images with BeanShell and creates a report of the cluster areas in the spreadsheet of Bio7:

http://bio7.org/flash/batchimagejr.htm

The script for the example can be found here!

Apropos: One reason that Beanshell and Groovy are used as default scripting languages in Bio7 is that they also accept (more or less) the Java syntax by default. Beanshell has the greatest Java compatibility. Some minor Java differences to Groovy are described here. Therefore it is very easy for a Java developer (or a C++ developer) at the beginning to use these scripting languages and if interested profit from the scripting syntax sugar these languages offer. In addition Java files can easily converted to scripting files and vice versa (e.g. for speed improvements).