Previous
Next
Finding significant functional groups in
clusters
From the
Attributes menu, choose Load > From File. Then select the
appropriate attributes file (.gxa). Attributes files are
available for download in
the Download section.
Analyze Attributes
From the analyze menu, choose Attributes. The above
menu will appear.
Here, you can select which of the loaded attributes
you wish to analyze. Usually, you will be interested in gene
attributes, as you will have loaded an annotations file that describes the
genes of your particular organism.
Minimum hits means that the attribute must be present
at least that many times in the cluster in order for the cluster to be
analyzed for enrichment of that attribute. This prevents, for
example, a small cluster with only one occurance of an attribute from
showing that that attribute is significant in the data.
Similarly, minimum # genes indicates that only
clusters of a certain minimun size will be analyzed. Again, this is
to prevent a cluster with extremely few genes from showing up as enriched
for attributes. A cluster that only has with 50%
of its genes having a certain attribute tends to
be much more meaningful in a cluster with 200 genes than in a
cluster with 2 genes.
Maximum p-value sets a threshold p-value. Only
p-values more significant that the selected p-value will register in
the analysis.
Pearson Cutoff is only available if your file is a TreeView format file, and your are analyzing "All Clusters".
This option allows you to set a Pearson correlation threshold, and it will return you a disjoint set of clusters created by walking
down each branch of the tree and finding the first cluster on the branch that whose Pearson correlation is above the threshold. If this cluster meets the
other criteria, it will be included. Otherwise, it will not be included. No other cluters on the branch will be searched.
walking down the hierarchy at each split cross this threshold
Spreadsheet vs. Graphical: Allows you to select whether you want to view your results quantitatively or graphically.
Click Analyze to begin.
Results
The program generates a table. For each attributes, we
see the following:
Cluster: The cluster in which the
attribute was detected.
P-val: The probability that the distribution of
the attribute in this cluster versus the dataset as a whole was due to
chance.
Hits: Number of times the attribute occurs in this
cluster.
Total: Number of times the attribute occurs in the
dataset.
Genome Total: Number of times the attribute occurs in the
genome.
#Genes, #Expts, mean, STD: Information about the
cluster.
The table is sortable by colums. Click on a header
to sort descending, and hit Shift+click to sort
ascending.
Click to continue:
- Next:
|
Find significant motif binding sites in your clusters |
- Previous:
|
Bookmark your favorite clusters |
- Home:
|
Tutorials Home |
| 
