To the BAR Homepage The Bio-Array Resource  eFP Browser - Development Homepage

Instructions for linking, uploading, or downloading the eFP browser.

Linking to the eFP browser dynamically

You can generate hyperlinks to the eFP browser such that when users click on them they will see the eFP representation for a gene that you're interested in. Create a link that looks like

http://www.bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi?dataSource=Developmental_Map&modeInput=Absolute&primaryGene=At1g01010 &secondaryGene=None&modeMask_low=None&modeMask_stddev=None

The CGI string options are as follows, whereby a | character demarcates the options.
  1. dataSource=Tissue_Specific | Abiotic_Stress | Biotic_Stress | Chemical | Hormone | Development_gcRMA | Developmental_Map | Seed. Default is Developmental_Map.
  2. modeInput=Absolute | Relative | Compare.
  3. primaryGene=AGI identifier | probeset identifier. This field is required.
  4. secondaryGene=AGI identifier | probeset identifier | None. Required in Compare mode.
  5. modeMask_low=None | on. For low expression level filtering in Relative mode.
  6. modeMask_stddev=None | on. For standard deviation filtering in Absolute mode.
  7. override=None | on. To activate use of a threshold value.
  8. threshold=value. To set a threshold value.

Uploading additional views to the Arabidopsis eFP Browser

Please contact Nicholas Provart for assistance with this. We will need the following:
  1. An image upon which the expression data will be painted. The input image is of type Targa (.tga), which can be saved at 32 bit resolution with compression. The regions to be filled are coloured with colours that correspond to the colours, in RGB hexadecimal format, e.g. #FFCC99, that are specified in the XML control file. We recommend using Photoshop to create this image and to colour the regions. Note that e.g. #FFCC99 is different from #FFCC98, even though your eye cannot tell these hues apart - useful if you want to use this image for other purposes. When flood filling regions, turn off the anti-aliasing option, so that colours are not dithered at the edge where the fill meets the outline. Example
  2. A copy of this image in .png format - this is displayed if no user input is supplied. Example.
  3. An XML file describing which samples are keyed by to which colour on the image, the image to use, linkouts and so on. See one XML file used by the eFP browser by clicking on the link.
  4. Gene expression data sets that we can load into the Bio-Array Resource. You can tell us where to find the appropriate data sets at NASCArrays, ArrayExpress or GEO. Ideally we would like the sets normalized via the GCOS method to a TGT value of 100, to facilitate cross-view comparison.
We are also receptive to adding other samples to existing views, let us know what you'd like to see added!

Downloading the eFP browser for local installation

Please contact Nicholas Provart for the source code under the GPL. Upon receiving the source code you will need to install it on a web server with several packages already installed. Please note that I cannot provide assistance with the installation of these third-party packages.
  1. Python, at least version 2.3, available from www.python.org.
  2. a modified Python Imaging Library (PIL), available from us. The Helvetica fonts are used by the eFP browser.
  3. The eFP browser makes use of a number of packages that may or may not be included in your particular installation of Python: os, xml.sax (saxutils), math, MySQLdb, cgi, tempfile, string and re. They are available from www.python.org.
  4. An SQL database containing gene expression data. In our own Bio-Array Resource database, three fields in a table containing many million expression values are used: data_signal contains the actual signal value, while data_bot_id and data_probeset_id are used to specify the appropriate signal to retrieve. The MySQL database engine is available from http://dev.mysql.com/downloads/.
In addition, you will need to provide and/or create the following files. I can provide assistance in creating these files.
  1. A lookup table or database describing which microarray elements map to which gene identifiers. The file should be a tab-delimited list of first the former then the latter.
  2. A database of annotations and gene aliases if annotations are desired in the output.
  3. An image upon which the expression data will be painted. The input image is of type Targa (.tga), which can be saved at 32 bit resolution with compression. The regions to be filled are coloured with colours that correspond to the colours, in RGB hexadecimal format, e.g. #FFCC99, that are specified in the XML control file. We recommend using Photoshop to create this image and to colour the regions. Note that e.g. #FFCC99 is different from #FFCC98, even though your eye cannot tell these hues apart. When flood filling regions, turn off the anti-aliasing option, so that colours are not dithered at the edge where the fill meets the outline.
  4. A copy of this image in .png format - this is displayed if no user input is supplied.
  5. An XML file describing which samples are keyed by to which colour on the image, the image to use, the database containing the gene expression information and so on. See one XML file used by the eFP browser by clicking on the link.
Finally, set up directories under Apache as follows.
  1. efp
  2. Under efp create a cgi-bin directory. The contents of the zipped file you receive are installed here. These files must be readable and executable by Apache, so chmod 755 them.
  3. Under cgi-bin, set up two directories called data and output. data must be readable by Apache and should contain the .png, .tga and .xml files as specified above, as well as the microarray element-to-gene identifier lookup file. output must be writeable by Apache processes, so chmod it appropriately. You will need to compile the modified PIL code in the Imaging-1.1.5 directory into a PIL directory under cgi-bin. The appropriate pilfonts directory should also be placed under the cgi-bin directory.
Then point your browser to your http://yourdomain.edu/efp/cgi-bin/efpWeb.cgi. Some tweaking of the code will be necessary to adapt the eFP browser to your local conditions, such as paths to images and the like. The output directory is cleaned up on a regular basis by the eFP browser itself, so no configuration of a cron job is necessary to remove older output image files.