When analyzing results from microarray experiments we encounter always the same problems :

• gene annotations in the result files are not complete or up-to-date
• combine different versions of hybridisation files
• filter out the "high quality" genes
• filter on data values which are derived from the hybridisation tables
• build expression matrices for clustering
• view simple analytical graphs from microarray data files

All this should happen with result files from our genome-wide yeast, mouse and human cDNA chips  ranging from 14000 to 55000 features. Some collaborators are using Filter-arrays with up to 200,000 features.
Some published data sets contain several hundred of data columns

Obvious solution might be to try MS Excel for these task. But this showed to become very time consuming, boring and error prone. Especially use of differing national settings (currency, date, decimal divider) may result in really nasty things happening automatically to your data. On the other hand Excel tends to crash when handling huge tables. Large data files (>65530 rows or >255 columns) can not be laoded completely anyhow.

Therefore, we implemented a tool which can do all the above described tasks : It serves us the nicely rearranged data tables like a butler : TableButler.

The program works on tab delimited files. Thus it is not really dependant on available computer RAM. Duzends (tested) or hundreds of files - even all hybridisation files on your hard disk - might be selected and processed in one go.

Available tasks :

• File selction : Source files are selected in a "File Open" like dialog one by one, or complete directory trees may be added.
• Table Merger : insert one table into an other via a link field (e.g. replace gene IDs by detailed gene annotations, useful for building ADF files)
• Table Rearranger : Remove data columns and resort order of data columns (useful to combine data from different image analysis programs, or to create shortened data files for publication).
• Row and Column filter : Filter genes which have minimal intensities or quality values for subsequent clustering.
• Recomput rows : Recompute new values from the selected data tables, e.g. invert ratios from dye-swaps or calc average intensities from red and green for filtering
• Recompute columns: Recompute data values in/from complete dolumns (e.g. global variance normalisation, ...)
• Build expression matrices : self explaining.
• Splice files : Combine files, or cut files into pieces.
• Data view: show various Scatterplots, histograms, line-graphs, heatmaps or boxplots.
• Edit : a simple spreadsheet editor allowing to edit really huge tables.

All tasks are configured interactively. The parameters for each task may be saved and recalled later.

A simple scripting language allows to set up and store complete processing pipelines. Scripts can be loaded and executed manually. The name of a script can be passed to TableButler via a command parameter, thus a Desktop Icon can be used to start a TableButler processing.
Furthermore, TableButler can be run as a server:  Several directories may be watched. Any TableButler script dropped into the watched directories is automatically loaded and processed.

You can modify TableButlers  appearance according to your preferences using command line parameters or inifiles

All functions may be adressed and configured from a graphical user interface. Previews of source and result files facilitate set up of filters and parameters with a view mouse clicks. Result files may replace the originals or saved separately.

TableButler was implemented with Borland's Delphi 5 and operates under MS Windows Win32, e.g. Windows 98se, Windows NT, Windows Millenium, Windows2000, WindowsXP.

Dowload of TableButler.