What is it that we’re doing with data generated by EyeWire?

Discovering neural networks! Here’s a behind-the-scenes glimpse at Jinseop’s exploration of new synaptic connections mapped by EyeWirers.

You’re familiar with the EyeWire overview.  How did we get here?

Before The EyeWire Games began, we identified over 100 putative synapses on J Cell #1, the first complete neuron mapped by EyeWirers (this is the cell that will be named by the winning team).  Below is an image of that process. In this and all additional images, purple is the J Cell and pink is the potential presynaptic cell.

Next, we put these seeds into EyeWire. Each team had 20 seeds at the beginning of the games. Below, the cell you see the pink cell in EyeWire, now blue. Players begin from this synapse and grow the cell out back toward the cell body.  This is how we begin to understand what type of cell is connected to the J Cell.

Teams play and grow their neurons. This generates 3D reconstructions, as you see below. The white segment is a piece of a branch which a player would see if he or she clicked “Start Playing.” Note that the color has changed but this is the same cell.

Below are screenshots and a vidoe of part of Jinseop’s analysis. You see a J Cell and a dendrite constructed by Team Facebook.

Take a close look at this awesome synapse:

Zooming out…

Finally, the whole cell view provides perspective on the complexity and size of neurons:

From this view, we can identify the type of the cell we reconstructed. This looks like a Starburst neuron, a large cell that makes inhibitory synapses onto J Cells. By repeating the process, we are trying to find out spatial distribution patterns of startburst inputs to the J cell, which will help explain how J cells function.