Table 6.

Cluster-specific over- and underrepresentation of GOslim terms

GOslim term	High	Low	Normal
cell wall			1,2,3
chloroplast	3	1	2
cytosol	3	1	2
ER			1,2,3
extracellular		3	1,2
Golgi apparatus			1,2,3
mitochondria	2	1	3
nucleus	1	2,3
other cellular components			1,2,3
other cytoplasmic components	2		1,3
other intracellular components			1,2,3
other membranes			1,2,3
plasma membrane			1,2,3
plastid	2,3	1
ribosome	2	1	3
unknown cellular components	1	2,3

Arabidopsis transcripts were clustered into three sequence clusters based on TIS similarity. Within these clusters, the numbers of gene models with associated GOslim terms are, respectively, 6,298, 5,039, and 3,019. Clusters denoted as being "High" for a specific term were determined to be enriched in genes labeled as such, relative to the full population of terms, by a randomization test. Those labeled "Low" were found to be relatively impoverished in genes labeled with the associated term, and "Normal" as being neither significantly over- nor underrepresented.