Optic Lobe Tag Rally

Looking at the optic lobe tagging dashboard Codex it looks like many/several of the cells that is labeled in our farms of completed cells are missing in the dashboard for instance:
TM1 L have 655 in dashboard and 678 in farm, and
Tm9 L have 219 in dashboard and farm in 722
What could be the reason behind this? tag not registered in database or are the dashboard based on a older version of the database from codex?

2 Likes

I know that there were a few questions/comments brought up in another thread so I’m moving the responses/discussion over here.

Re: Differences in numbers between Flyer “Farms” in the Q&A Log and the Codex Tag Dashboard.

  • Most of the numbers are relatively close (+/- 20) to the numbers listed in the farms. This could be a difference in the parameters for the query run in Codex (it is based off of similarity in morphology, synaptic connectivity, input/output neuropils, etc.) to create the tagged/candidate lists. It could also mean that maybe some cells were prev. mislabeled or that the query missed some cells b/c they didn’t line up with the parameters. This is why we have the candidate lists, to help us find cells that have missing labels, may have gotten missed previously, and/or to correct any mislabels.

  • Codex is versioned data so it is possible that some cells have been updated since the “% tagged” lists were generated. The retinula cells (R1-6 & R7/8) are a prime example of a list that needs to be updated since the last Codex version at the end of April – there should be a great many more in the tagged list!

Re: Candidate list updating

  • While the lightbulb status in FlyWire updates within a few minutes of submitting a label/tag, please be aware that Codex only updates approx. every 24 hours with the tags.

Re: Preventing Overlap of Tagging

  • The Tag Dashboard is a beta version and we are developing methods to help prevent users from checking the same batch of listed cells (Update here!)

  • Checking the lightbulb status is the most efficient way to confirm labels; but please be aware that some candidates may be in the list b/c they might be mislabeled! You can filter for unlabeled candidates on the list by adding {and} {not} label to end of the text in the search bar and hitting search.

  • I recommend tagging the (now) left side of the Optic Lobes. Unsurprisingly, it is the side with the lowest percentage of tags. It is also the area with the least overlap with other users.

Re: Dm8s

  • A Princeton lab tracer is currently focusing on the Dm8s
2 Likes

As for the tagging in the left lobe. I want to do at least some more Tm and TmY in the right lobe and only then move to the left one. I hope, that’s ok.

1 Like

Perfectly fine! I’m not aware of anyone in the lab tagging any Tm or TmY cells, so they’re fair game :slight_smile:

1 Like

An FYI that I am gardening and tagging the (now left) Lobula Plate. Primarily focused on the TM4/TM5 cells right now but tagging others as I come across them. I haven’t updated my farms in the Google Sheet yet but so everyone knows where I’m actively working.

3 Likes

I’m working my way through the L1-5’s in the (new) left lobe atm, starting with 1’s. Keeping farm links up to date as i go. Will go back to the right lobe after to catch any that were missed.

3 Likes

Thanks all for posting and joining in on the tag party! :partying_face:

@st0ck53y No need to worry too much about the L1-5s on the (now) right OL. In the lab, twister (yes, originally from Eyewire!) is going through to catch any missed ones on that side.

2 Likes

Update: Candidate lists on Codex Tag Dashboard will now randomly shuffle

To prevent everyone from checking the same first 10 cells in a candidate list, the list order of the cells will shuffle every time you open a new instance of the candidate list for a cell type. This is great if you’d like to just pop in and tag a few cells for a short period of time :slight_smile:

If you’d like to keep a stable candidate list, you can download the Cell IDs text list.

After a cell is labeled, it will be removed from the candidate cell list when Codex is updated (~ once daily).

1 Like