...

You have the choice to use the Fisher's Exact or Chi-Square test. Both tests compare the proportion of a gene list in a functional group to the proportion of genes in the background for that group. Both are acceptable and you can always test both by re-running the analysis. You can also restrict the analysis to functional groups with more than or fewer than a specified number of genes. Restricting the analysis to GO groups with fewer than 150-200 genes will increase the speed of analysis and exclude large groups which may not be too informative. If analysis time is not a concern, you can just use the default settings.

...

The new spreadsheet (GO Enrichement.txt) is a child spreadsheet of the gene list. The first column contains the GO functional groupgroups, each of which falls into the broader categories of biological process, cellular component or molecular processes shown in column 2. The GO functional groups are arranged by descending enrichment score, which is shown in the third column. The enrichment score is the negative natural logarithm of the enrichment p-value, which is shown in column 4. The higher the enrichments enrichment score, the more over represented this functional group is in  a functional group is in the gene list. As a rule of thumb, if a functional group has an enrichment score of over 1, it is over represented. A value of 3 corresponds to significant over representation (p-value=0.05). For your data, you may wish to add a multiple test correction (e.g. FDR) by going to Stat > Multiple Test correction. We will not perform the multiple test correction for this tutorial.

There is more information present in the spreadsheet which helps describe the enrichment score, including the percentage of genes in the group that is present in the gene list, the number of genes present in the group that are present in the list and the total number of genes in the group. Because the original gene list was derived from statistical analysis, extra columns will appear for all p-values in the ANOVA model. For example, the Young/Old score and Gender score columns contain the negative natural logarithm of the geometric mean of p-values for each marker/gene present in the list and in the group. These scores represent the level of differential expression of the genes in the functional group. The larger the score, the more differentially experessed the genes are in the group. A score of 3 or greater corresponds to an average p-value of 0.05 or less. For example, the Young/Old score explains how differentially expressed the genes present in the list and in a given group are between the "Young" and "Old" categories.