Home » forums » LtU Forum

visual programming of R: Red-R

via PPIG mailing list, presumably some of you have already seen the email; i didn't find any hits for Red-R searching LtU.

Red-R makes the advanced functionality of R available to the
non-computational users by hiding the computational complexity behind
a visual programming interface. In addition, Red-R improves analysis
readability and data sharing to facilitates better communication in
inter-disciplinary teams.

Analyses are performed by visually linking a series of widgets
together that read, manipulate, and interactively display data. These
pipelines, representing both the data and analysis, can be easily
shared with others. Red-R can also generate reports in odt, html and
latex to help better document and share results.

By raould at 2010-11-02 17:48 | LtU Forum | previous forum topic | next forum topic | other blogs | 7871 reads

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

competitor

kinme says "From day one, KNIME has been developed using rigorous software engineering practices" which i certainly guess is a dig at R.

overall, just curious to think about how much visual programming can/not open up the field of statistics.

By raould at Tue, 2010-11-02 20:22 | login or register to post comments

Visionary programming for statistics

If you want visionaries, read Bradley Effron, the inventor of the bootstrap. The appendix of his Springer-Verlag book is a manifesto where he basically argues for drag and drop statistics environments where the user does not have to really know much beyond what estimators of a distribution are, and which ones they should be interested in.

By Z-Bo at Tue, 2010-11-02 20:39 | login or register to post comments

ok

hope to get to do that (checking my interlibrary loan).

in my mind's eye, i can see various Star Trekkian / HAL uis for statistics that could be pretty nifty. if the system knew enough about the data it could put up red flags, green flags, to help guide what is statistically reasonable to do to the data.

By raould at Tue, 2010-11-02 21:15 | login or register to post comments

motivation behind Red-R

No visual interface will be flexible enough to replace the command line for those of us that know programming/scripting. On the other hand, no matter how powerful a command line interface is, for those that don't know programming, its very unlikely they will learn and use it.

Red-R was initiated while I was working on a graduate degree in computational biology and needed a way of communicating my work with others that had little or no computational background.

Red-R was really created to bridge the gap between these two groups. Do your work in R. Bring the R session into Red-R and create pipeline around your work that will allow others to interact with the data and results without requiring knowledge of R programming.

By anup at Wed, 2010-11-03 15:25 | login or register to post comments

That's interesting. Can you

That's interesting. Can you say more? What kinds of analysis? What kinds of context did you give these presentations?

By Ehud Lamm at Thu, 2010-11-04 06:01 | login or register to post comments

example analysis in Red-R

Large dataset analysis has become common place in biology. In a typical project a biologist will perform an experiment that results in millions of data points. A bioinformatist will take the data preprocess/normalize the data and do some data mining to find patterns of interest. At this point the real problem is what's interesting. Usually the biologist has a much better understanding of the system and so really they should be looking at the data to make conclusions.

In Red-R the bioinformatist can preform some preprocessing analysis (either in R or Red-R) connect a few widgets to these data to sort/filter and finally visual the results (tables/graphs). Now a biologist can start interacting with Red-R to change the sorting/filtering or add different visualizations. If the preprocessing was done in Red-R the biologist has all the parameters in a GUI and can alter these to really start exploring the data and analysis to see how the results change. In my experience this ability to interact led to better data interpretation and conclusions.

We have used this tool for mining microarray data. This involved data normalization/scaling, building a regression model, identifying genes that behave differently under experimental conditions and then looking for enrichment of specific biological pathways.

We are hoping that as some of these projects are published, the whole analysis will be shared as a Red-R file for better reproducibility and readability.

By anup at Thu, 2010-11-04 18:12 | login or register to post comments

Thanks. I was interested in

Thanks. I was interested in specifics, i.e, more details on what you describe thus: We have used this tool for mining microarray data. This involved data normalization/scaling, building a regression model, identifying genes that behave differently under experimental conditions and then looking for enrichment of specific biological pathways.

By Ehud Lamm at Thu, 2010-11-04 22:17 | login or register to post comments

current features

Got it.

Basic functionality:
Read/View Data
Subsetting
Merge/Intersect/Filter
basic math functionality
Plotting: lines, histogram, boxplot, interactive scatter plot ...
Stats:ANOVA, LM, t-test,f-test...

Specific functionality:
Bioconductor microarray analysis
Survival analysis
SQLite
ROCR – ROC Curves
Neural Nets
LME4

work in progress:
Spatial Stats
Econometrics
Multivariate Distribution

By anup at Fri, 2010-11-05 00:10 | login or register to post comments