Why cluster the transcriptome?

Question

[Efejir]

This is an application question for a biochemist. What extra information can clustering RNA-seq data give about the particular state of the cell at a given time? For instance, micro arrays tells us that some genes are activated in response to a particular stimulus. We deduce thus, what pathways are active (and the ones that are not) from this. Likewise, measuring mRNA quantities in RNA-seq tells us that at a particular time, so and so mRNA is produced in large amounts, an indication that, at that time, a pathway may or may not be active adjudge from the protein constituted - pretty much the same conclusion a micro array reaches eventually. I can tell this just looking at the quantities of mRNA quantities (using a filter of some sort I think) which brings me to question why I need to cluster at all.

Cluster can reveal hidden relationships. Great! But of what biological importance are these relationships and hence clustering?

Answer 1

[agd27]

The answer to this question is highly context-dependent: i.e., what values are you clustering, specifically? Clustering the expression values from a single experiment, for instance, will only tell you about which genes are correlated in expression value. As per your observation, this probably does not say much about the biological importance of these correlations. However, clustering expression fold-changes between cells before and after treatment with a drug, for instance, can tell you a great deal about which genes are coordinately regulated, which is (in general) very biologically meaningful. As with all data analysis methods, it's all about the interpretation.

[Efejir]

I really appreciate your response agd27!