1-Photon 2-Photon Ratio Data

[randismaael]

Added 1-Photon 2-Photon data for more opsins in notebook, updated the default 1p/2p ratio in opsin library based on the averages of the opsin data.

[kjohnsen]

Looking good, but needs to be cleaned up a little more. Mainly clean up the notebook so a reader can follow how we got the ratio numbers. I am myself confused over the full vs. first-two-points slopes, since the notebook makes the latter look better, but use the former for ratio calculations.

The rest is fairly minor.

If you want to take a stab at fixing the failing tests, that would be awesome! That would probably entail cloning the tklfp repo and updating how it's doing scipy interpolation

[kjohnsen]

we should probably get rid of this whole nonlinearity part, since we ended up deciding we want just the linear model. We want this to serve as documentation of how we ended up with the numbers we did, so this is a distraction

[kjohnsen]

there's a lot of code repeated for each opsin. better to make a function or use a loop to not repeat yourself

[kjohnsen]

again, I'd restructure code to use a loop or a repeated function to not have to repeat the same thing in a bunch of cells

[kjohnsen]

An explanation is missing here of why we switched to using the slopes from the first two points instead.

[kjohnsen]

again, refactor to not have code repeated per opsin

[kjohnsen]

Why did we use the full-data slope (instead of first-2) for the ratio calculations? What is the takeaway from these first2 plots? Which do we want to go with and why? The first-2 graphs look good to me, but it would be good to show the full-data slopes to compare.

[kjohnsen]

Looks great! Just saw a mismatch in the ratio. And we need to get the tests passing.

[kjohnsen]

Just need to update all-optical tutorial

[kjohnsen]

Sorry, this will be a bit of a tricky change. Need to correctly model how 2p activation scales quadratically (since each activation requires 2 photons to coincide). This will require changing the Brian model itself of light and/or opsins and recalculating the ratio.

[kjohnsen]

We generally don't want comments that only explain previous code

[kjohnsen]

Nitpick: Extracted would be extracted (lowercase), in middle of sentence

[kjohnsen]

This is going to have to be recalculated with the quadratic scaling, inferring activation from photocurrent, etc.

[kjohnsen]

As I learned a bunch of stuff from AI I had glossed over, realizing activation should scale with $P^2$ for 2p opto, and seeing that the data we were using here did not scale quadratically, I realized we should probably use simulations to learn the map from $\phi$ to photocurrent, so then we can map backwards from photocurrent to $\phi$, keeping in mind that $\phi \propto P^2$ for 2p

TLDR: we need to recalculate the ratio 🥲

More details towards the end of this Gemini chat