Perform Principal Component Analysis on Agilent Two Color Dataset

[Miserlou]

Context

We have the data!

Now we need to know: is the data good? Specifically - Can we clearly separate channel one from channel two? Even better - can we automatically classify an experiment as being a reference or loop experiment?

Problem or idea

Solution or next step

Rich relearns how to use Pandas, scikit-learn and Jupyter.

[jaclyn-taroni]

Related: #87 (where the original questions were posted)

[Miserlou]

The GSE that are processed in that file are:

GSE35576
GSE22900
GSE15109
GSE23669
GSE38242
GSE28748
GSE21367
GSE39477
GSE30181
GSE68081
GSE45403
GSE58295
GSE42401
GSE38241
GSE26129
GSE25346
GSE19324
GSE29917
GSE55668
GSE51081

[35576, 22900, 15109, 23669, 38242, 28748, 21367, 39477, 30181, 68081, 45403, 58295, 42401, 38241, 26129, 25346, 19324, 29917, 55668, 51081]

[Miserlou]

I did a thing

[Miserlou]

[jaclyn-taroni]

@Miserlou can we put the notebook you used for this over in https://github.com/AlexsLemonade/agilent-two-color (via PR)? I think that's a better spot for review and will be helpful if we 1) have multiple folks working on this (you & me most likely) and 2) if we take this further -- i.e., automatic detection of different designs.

[Miserlou]

Yeah for now it's been here: https://github.com/Miserlou/Science

[jaclyn-taroni]

Whatta name!