Initial clustering on continuous light data

[murraycutforth]

I would like to carry out an initial exploration of clustering on a handful of manually extracted features (following on from #14 ). This will involve preparing data in the form of a table, where each row corresponds to a mutant and each column corresponds to a feature, and then applying a bunch of off the shelf clustering methods (k-means and GMM would be the default methods to try).

To keep things simple, I will initially look at just 20H ML and 20H HL data.

This approach can also be sanity checked by first making a table where each plate-well combo has it's own row, and seeing if the two light treatments separate into two clusters. Then I'll proceed to using one row per mutant and check that all the WT colonies cluster together.