Alternative controls

[deeenes]

Give option to choose one or several controls to compare drug treatments to

[deeenes]

This is possible today by multiple runs, isn't it?

[deeenes]

Now the baseline control can be any set of compounds. For example, if we have A, B and C compounds, {"A", "B", "C"} means the combinations A + B, A + C, B + C will be considered as control.

But it can not be, let's say, only A + B and A + C, but not B + C. Let me know if such feature is necessary.

[NgocHien-Du]

Below is the situation with several controls:
Control 1: A + B + a1 + a1
Control 2: A + B + a1 + b1
Control 3: A + B + b1 + b1

The treatment could be:
Treatment 1: A + B + d1 + a1, then compare to Control 1 (because d1 is in similar setting as a1)
Treatment 2: A + B + d2 + a1, then compare to Control 2 (because d2 is in similar setting as b1)
Treatment 3: A + B + d1 + d4, then compare to Control 1 (because d1 and d4 are in similar setting as a1)
Treatment 4: A + B + d1 + d5, then compare to Control 2 (because d1 and d5 are in similar setting as a1 and b1, respectively)
Treatment 5: A + B + d2 + d4, then compare to Control 2 (because d2 and d4 are in similar setting as b1 and a1, respectively)
Treatment 6: A + B + d2 + d5, then compare to Control 3 (because d2 and d5 are in similar setting as b1)

It can be increased to have more than 3 controls. But probably we should just start with this example.

It could be solved by e.g. the user specify for each treatment what control to compare to (by providing a table of all treatment with their corresponding controls in the config).
Or the user specify for each drug (d1, d2, d4, d5 in this example) what are their corresponding similar settings (a1, b1 in this example) and the program compute the corresponding controls.

It would also be nice to have an option in the config whether there are several controls to compare to or not, because there are times when we only have one control.

Let me know if it is clear.

[deeenes]

No, it's not clear 🙉

What is a1, b1 and d1-5? And what is "similar setting"? In my example I talked about compounds and called them A, B and C. You're saying d1-5 are compounds. Apparently A and B are assay reagents and cells, in your example? And a1 and b1 are the plain medium channels? And those are different?

Anyway, it's certainly possible to do this, though I need some more clarifications from you. With such controls, the statistical power will be much lower, as we have around 100 medium only plugs in one experiment, while a single sample consists of about 10 plugs.

[NgocHien-Du]

Sorry for my bad explanation...

A and B are assay reagents and cells.

a1 and b1 are the medium that does not contain compound, but the solution used to dissolve the compound. Not all compounds are soluble in the same solution. Some are soluble in water, some are in DMSO, some are in ethanol. Therefore, as a control, we have to add the same amount of one of these solutions (water, or DMSO, or ethanol) to the plain medium. We know that addition of these solutions (and their concentration) can stress the cells, therefore change the cells' responses to drug.

d1-5 are compounds.

For statistical power, if needed, we can produce more plugs with these controls to increase the power.

I hope it is better...

[deeenes]

Aha, I understand. Yes, Vida too asked for DMSO control, this is quite standard, but then do you work with different DMSO/Et concentrations for some compounds? Alright, that's no problem, we can have this kind of controls :)