The human anatomy dataset made available in this resource consists of individual tissues for which the average expression vector was created from at least 34 samples. The tissue types were chosen such as to exhibit a high diversity, while at the same time the number of similar tissues was minimized. The dataset therefore comprises a robust expression matrix across a diverse but minimally redundant representation of the human anatomy.

In a first step, we performed a Principle Component Analysis to cluster the samples and identify groups with similar biological function. The plots for the first three principle components are shown below.

legend for PCA plots

component 1 vs 2
Component 1 versus 2

component 1 vs 3

Component 1 versus 3

component 2 vs 3

Component 2 versus 3

All three plots clearly separate individual clusters from the remaining tissues. For example, the central nervous system tissues show a common but very distinctive profile relative to the other tissue types. In a second step, we therefore searched for genes that are differentially expressed between an individual cluster and the remaining tissue types. Here, we show the example of the CNS cluster versus all other tissue types. This was performed using the RankProd package of Bioconductor by comparing these two groups and identifying the most significant genes. The top 50 genes are listed below.

CNS specific genes

Top 50 most differentially expressed genes in the comparison Central Nervous System versus all other tissues, as calculated with the RankProd package from Bioconductor. Listed are genes expressed in CNS but not expressed in other tissues.