Tree Clustering; overviewing groups of genes with similar patterns.
Tree clustering (hierarchical clustering) is useful to group genes by expression patterns.
00:00 Before you apply Tree Clustering, you need to exclude noise, which are
- unreliable signals.
- unvarying signals.
01:55 The samples are largely divided into normal and tumor group.
Let's extract genes which are highly expressing in normal and low in tumor.
02:20 The starting level varies among patients, but the trend is downward in tumor.
If you extract genes sharing a certain pattern, it's reasonable to see their expression levels.
02:50 Apply clustering again, but on "Cluster 1" list.
No normalization block for making ratio is applied, select "Euclidian" in Similarity Measure.
Genes sharing a same pattern were divided into some clusters by expression level.
03:55 Run PCA to look the data from another angle.
PC1 looks reflecting the difference between normal and tumor, and PC2 looks separating the effect of tumor.
04:15 Extract genes contributing to PC2.
04:45 Apply clustering on the list of PC2 contributing.
05:00 Now you see the groups of tumor samples, and their featuring genes.
Please request a free Online Support, if you don't know how to use this tool.