RReportGenerator English

From Wikili
Jump to: navigation, search

RReportGenerator provides a 'report-generator' tool for routine automatic statistical analysis with "R" in a highly user-friendly way via a graphical user interface (GUI) or an interative web page RReportGenerator on the Web

While the statistical platform "R", see also the official site of R, and the vast collection of additional modules on CRAN and Bioconductor allow very powerful statistical analysis, it’s particular command-line syntax renders the program difficult to access for non-statisticians. In this context we have developed a program designed for routine executing of a predefined "analysis scenarios" for a given problem which can be easily operated by non-experienced users via it’s graphical interface. In turn a pdf-report with the analysis results, tables and figures is generated that can be accompanied by supplemental data-sets for export to other programs (e.g. Excel).

Analysis scenarios are written in the R and Latex language and allow following a path of multiple steps of data-treatment including flexible generation of graphs and potential identification of warnings. The graphical user interface of RReportGenerator allows the user to simply choose among predefined analysis-scenarios to be applied this to a given data-set. In turn a pdf-report with the analysis results, tables and figures is generated that can be accompanied by supplemental data-sets for export to other programs (e.g. Excel). An example of mouse retina transcription profiling illustrates suitable tasks of quality control and analysis for automated analysis.

The program RReportGenerator (compiled for Windows and Linux) and further informations, tutorial and examples are available on Wolfgang's site
Besides, we're also developing a web-version RReportGenerator_on_the_Web allowing to run calculations on separate serveurs.

Applications :
Predefined analysis-scenarios for automatic analysis have been developed for the following areas :

  1. Transcriptomics : Analysis scenarios combining a large collection of different types of quality control (QC) for Affymetrix genes expression chips .
    There is also a scenario for QC of printed arrays (images analyzed by MAIA).
  2. Comparative genomic hybridization, CGH:
    The available scenarios use 3 or 4 different algorithms of segmentation and allow supperposing the results.
  3. Transfected cell array (TCA) : The analysis scenario(s) may be used to define a threshold based on non-treated control samples and to report a summary about all samples tested on a transfection plate.

Please check the supplemental information available with the analysis scenarios available through the www-library in RReportGenerator for further details. Using the www-library directly accesses the most recent versions of our public analysis scenarios.

In conclusion, RReportGenerator allows to run routine statistical analysis while benefitting from the environement of R via a convenient graphical interface (GUI) allowing inexperienced users to run routine analysis tasks.

Reference:
The program is published and accessible as open access : Raffelsberger W, Krause Y, Mouliner L, Kieffer D, Morand AL, Brino L, Poch O;
RReportGenerator : Automatic reports from routine statistical analysis using R. Bioinformatics 2008, 24(2), 276-278

Limitations:
The overal reliability of automatic analysis depends very much on the capacity and flexibility of the algorthims used to recognize (and adopt data analysis accordingly) for special cases. In practice this means that the interface human experimenter to machine (and the other way round) remains a delicate place where numerous "misinterpretations" or "misunderstanding" may happen. In this context the analysis scenarios implemented with RReportGenerator don't intend to replace in-depth analysis performed by a real specialist, but rather to aid such analysis by providing those elements that can be easily generated using automated procedures.
Physical/compoutatinal limitations:
Treatment of larger numbers of Affymetrix gene-exporession profiling arrays requires large amounts of RAM on your computer (see also the numerous discussions in the R news-groups). On a PC with Windows XP with 1 GB RAM you can treat up to 30 Affy arrays with the analysis scenario 'automAffyQC1.Rnw'. Up to 200 Affy arrays have been treated sucessfully on a Linux server with 16 GB RAM, further improvements (in particular for the time requiresd) for large jobs are under development.



... voir cette page en Francais : RReportGenerator