Getting Started: Comparing two experiments

Once GOALIE has computed the relationships among different windowed clusters, it can use this information to perform a high level experiment comparison. To do so we need to load a new experiment through the Main GOALIE Interface. Bring it forward and load a new windowed clusters experiment.

The Main GOALIE Interface

The Main GOALIE Interface is the actual entry point for the GOALIE system. The main interface allows you to do several (maybe too many) things. It has the following functions.

• It organizes the windowed clusters experiments, it displays bits and pieces of them and it can run the various enrichment re-description methods that GOALIE provides.
• It is capable of loading single clusters and organize them into windowed clusters experiments.
• Eventually, it will be able to load "raw" data in the workspace and it will allow a user to set up its own clustering experiments.
• It can browse several terms structures (i.e., controlled vocabularies and/or ontologies)

For the sole purpose of experiment comparison, let's now get to the File->Open menu. Choosing this menu item will pop up the following dialog asking you what kind of "file" you want to load in the workspace.

Let's choose Complete Cluster Set Script and select (in the following file chooser dialog) the file

GOALIE will load the script file and the relative clusters that make up the complete windowed clusters experiment. The final result will be to have a second line in the list of windowed experiments.

Let' select the second experiment (the one just loaded) named Spellman Yeast Cell Cycle Data - Elutriation. We now want to run the enrichment re-description algorithm on such an experiment. We do so by selecting the Fischer Exact Test method in the Time Course Computations panel. This will bring up a sub-panel asking for one parameter (in this particular case).

Let's now just hit the Run method and wait (again, possibly for tens of minutes depending on your machine and configuration; GOALIE may seem unresponsive, but it is not).

After the computation finishes, GOALIE will pop up a new window with all the cluster relationships information relative to the elutriation experiment. This window is just like the one previously illustrated. You can browse trhough the information collected and see whether there is anything in the experiment that is of interest to you.

But let's move on to produce a view of the possible similarities between the two experiments. Go back to the Main GOALIE Interface and select the Tools menu. Let's choose the Compare windowed clustering experiments menu item.

Such a choice will pop up another small dialog where you can choose two experiments for comparison. Let's choose both experiments. and hit Ok.

GOALIE will now run an alignment program on the summaries profiles of the terms used in the enrichment re-description. The result will be displayed in a new window showing the aligned profiles in order of similarity.

Browsing through it we can see what are the similarities of the two experiments with respect to (in this case) the enrichment re-description done in terms of the GO process ontology.

• Back to Getting Started with GOALIE.
• Back to GOALIE Experiment View.