Submitting Synapses

I think we’ve all seen the ‘synapses with partner’ button in the lightbulb, and if im not mistaken @Amy_R_Sterling has said we (players not researchers/admins) can use it, and I would like to but I don’t want to make a mistake so I have a couple of questions:

  1. Sign in - Google Accounts in this link what is the cleft score threshold? By default it is at 50, is that something I/we should know to adjust based on ‘parameters’ or just always leave it at 50?

  2. How important is adding the date and time? I mean its not difficult or anything and UTC is just -2 hrs from my timezone, -1 hour from central/western europe time, just curious if its mandatory or do it if u dont mind.

  3. What do synapses look like? I mean i know the easier ones, the bouton wrapping on bouton/nub, like this:
    Screenshot 2022-08-10 102538

But im not quite sure when i see a bouton wrapping itself around a branch or 2 nubs touching if those -also- constitute a synapse or not, like this:

(the small pink circle surrounded by the red on 3 sides).

I would like to contribute and help submit synapses but i’m afraid of submitting the wrong thing(s) as synapses, I’ve submitted 2-3 very easy (like the 1st image above), but perhaps other players and/or admin(s) can give various different examples of synapses (screenshots) here so that any player who wants to submit synapsing partners can.

Finally, perhaps this should be a thread/topic of its own but what do all the things mean in the connectivity option in the lightbulb? Sign in - Google Accounts

There’s a lot of info in there and it all looks very interesting but ofc I haven’t the slightest clue what any of it means, if an admin doesn’t mind spending 5 mins to explain the basics i’d greatly appreciate it.

3 Likes

The easiest way to spot a synapse is to look for postsynaptic densities.

On the picture above these are most probably synapses. There are probably others, but I’m less sure about them.

There are also electric synapses, which I don’t know, what they look like. There albo might be synapses parallel to the slides, so harder to decide, if we see a cell wall or a postsynaptic density.

The invaginated one, you’ve shown on the previous picture is not necessarilly a synapse accorcding to this article: Frontiers | Invaginating Structures in Mammalian Synapses

As for the rest of the questions - I have to idea :smiley:

2 Likes

oh interesting, according to that article one of the most common source of mergers are synapses


ive very often seen the red circled merged in FW (and come to think of it in ew as well, all those fused boutons).

2 Likes

Indeed, interesting. Probably because the segmenting or clustering algorithm isn’t sure, where’s the wall is or if the wall even exists and probably treats it as some sort of an organelle.

2 Likes

i had the impression that the ai mostly found the synapses, but of course there have to be mistakes there also. Also you will not be able to look after synapses when tracing but have to first add the layer showing synapses
https://ngl.flywire.ai/?json_url=https://globalv1.flywire-daf.com/nglstate/6189628956606464


from a uncompleted branch

if we are supposed too look for synapses and/or identifing cells we will need a lot more instruction on how to do it accuratly, and maybe some training in the sandbox before beeing let lose in the dataset

4 Likes

oh i didnt know an/the AI found synapses as well, how do I add those layers in other segments? Idk if we’re ‘supposed’ to but maybe able to? Yeah I guess there’d have to be some kind of training and/or reading material. Still it could be interesting, finish a cell, declare it complete (axon, dendrites and CB all there etc) then spend some time (if wanted) finding the synapses the AI missed, I could dig that.

1 Like

You find the synapses going lightbulb- connectivity- generate ng link with all synapses (blue button at the end of page) this will only show the synapses on the cell you have chosen and not any new segments you add.

3 Likes

You are more than welcome to use the Synapse with Partners tool! It lets a user select 2 tools then find all the synapses between them, so you do need to have cells on your screen that are connected. What you may find more interesting is the Connectivity tool, which lets you query all up/downstream synaptic partners of any cell and then generate a mesh with all of those.

  1. You should not need to adjust cleft score. I’ve asked Sven exactly what the score is and will report back once I have an answer. Edit @Nseraf here’s the answer I got for “what is cleft score” Sign in - Google Accounts "“cleft_score” contains the scores acquired by Buhmann et al. by using the synapse segmentation from Heinrich et al. [2] (higher is more confident). For the latter, Buhmann et al. found the highest value in Heinrich et al.’s segmentation along the line between the pre- and postsynaptic point for each synapse. This represents an individual assessment of the validity of each synapse because Buhmann et al. found synapses without using the segmentation by Heinrich et al… Buhmann et al. suggested using the product of both scores, others found that the “cleft_score” alone with a threshold of 50-100 to be a good filter. "

  2. No need to input date/time

  3. I’ll see if I can put together some examples of synapses.

2 Likes

Good idea. Let me put together a blog walking through these items!

4 Likes

Thanks for the reply!! Sounds like a plan :smiley: complete a cell then do the thingy hehe and thanks for the info!

2 Likes

@Nseraf @Krzysztof_Kruk @annkri

Here’s a guide that Szi-chieh put together overviewing synapses in fly. “The internal links are not working anymore because we lost our entire wiki during an update but external links should still work. Fly is a bit different from zfish and mouse. The PSD is not that easy to see sometimes. That’s why there is a special definition for fly synapses that we follow.”

5 Likes

This is quite interesting and informative, and the part where I start reading it then I have questions, you/other admin get answers from author/researcher(s), then the answers beget more questions and so on and so forth and after 3 years of this I get a neurology degree from princeton :stuck_out_tongue: (if only lol)

Jokes aside, thanks!

4 Likes

Yeah, I feel lately, like I’m on a university again with how much new things we learn, how we work as a community, how good teachers/admins we have and how fascinating our own research is :slight_smile:

4 Likes

hah ikr? I’m -also- wondering assuming that the community start doing this, submitting synapse partners/synapses how much work it’d be to add that to our profile stats in-game, ie: Synapses ## (assuming that submitted partners is something that is counting in the background (after admin/researcher reviews etc).

3 Likes

… and number of identified cells

4 Likes

Yes, This will def be in profiles! I don’t have a good ETA on when that will happen but here is a WIP (very work in progress - the final design should look better, more like cell complete notif in EW) but it includes the expanded stats that Kai’s working on in the backend. There will be some badge / icon system but it isn’t made yet.

“Princeton University” would prob be replaced with “Location” and hopefully there will be a way to highlight some favorite cells that you have completed

5 Likes

That already looks great :slight_smile:

1 Like

that looks fab!! Can’t wait :smiley:

1 Like

Love this! I shared it and the one above from @Nseraf with the broader FW team. Our journey here mirrors the journey of science itself! Every good question leads to at least two more. Looking forward to the next years of connectomics with y’all!

4 Likes

Question, if you know (the answer):

the expanded stats kai is currently working to in our profiles, will they grandfather anything we do now? ie: say I submit 10 cells type identifications today and admin/researcher accepts/admits them also today, and kai releases the new expanded stats tmrw, will my profile’s Types say 0 or 10?

thnx

2 Likes