attachments

[Sanja Jasek]

Hello,

I have several questions about attachment connectors:

1. We want to mark attachments like tight junctions and desmosomes. What would be the best approach for this? We think the attachment connector type is closest to what we are looking for, but that's only one connector type and we want to mark several different junction types.

2. There are two ways of adding the attachment connector, with different color codes:

a) select a skeleton, alt+click somewhere and select attachment, shift+click on another skeleton (upper connector in attachments_2018-10-37.png)

b) select a skeleton, alt+click somewhere and select attachment, alt+click on another skeleton (lower connector in attachments_2018-10-37.png)

This is confusing. Does the b) version have a function?

3. Only attachments marked with method 2.a) show up in the connectivity widget. Is this intentional?

3. When listing connections in the connectivity widget, attachments only show up on the "pre-synaptic" neuron (attachments_2018-10-37.png shows the "post-synaptic" neuron). But aren't attachments supposed to be non-directional?

4. Setting attachment as the default connector type doesn't work and I get a warning (see unknown_connector_type_2018-10-31.png). This only happens with the attachment connector type; synaptic, abutting, and gap junction all work.

Best regards,

Sanja

[Tom Kazimiers]

Hi Sanja,

On Wed, Oct 31, 2018 at 04:26:47AM -0700, Sanja Jasek wrote:
>I have several questions about attachment connectors:

Just to provide some context, attachment connectors were originally
created to link vesicles to a particular synapse or location on a
skeleton. The relevant relations with respect to a connector are
"attachment_of" and "close_to". A attachment connector is attached to a
particular treenode. Other single treenodes can then link to it using
the close_to relation.

>1. We want to mark attachments like tight junctions and desmosomes.
>What would be the best approach for this? We think the attachment
>connector type is closest to what we are looking for, but that's only
>one connector type and we want to mark several different junction
>types.

I guess it depends a bit on how exactly you want to annotate this. If
you want to attach the mere fact that there are tight junctions and
desmosomes at a particular location in a skeleton, an attachment
connector with its close_to relation would only work if you woul
additionally qualify the attachment treenodes with an annotation or,
alternatively, tag the connector nodes.

Looking at your screenshot though it looks like you would want to link
two skeletons through such a junction connector or desmosomes connector.
In that case I believe it might be better if I would just add these
connector types to CATMAID itself. I suppose both types would work like
gap junctions, i.e. non-directional and a maximum of two skeletons can
be linked with a gap junction connector? This would make it easier to
look at the two types separately from each other. It would be relatively
easy to add support in the creation context menu, default connector
types, Connectivtiy Widget, Graph Widget and 3D Viewer.

>2. There are two ways of adding the attachment connector, with
>different color codes:
>a) select a skeleton, alt+click somewhere and select attachment,
>shift+click on another skeleton (upper connector in
>attachments_2018-10-37.png)
>b) select a skeleton, alt+click somewhere and select attachment, alt+click
>on another skeleton (lower connector in attachments_2018-10-37.png)
>This is confusing. Does the b) version have a function?

I agree, this behavior is confusing. Alt + Click has two modes of
operation at the moment, depending on whether a connector or a treenode
is selected. Originally, Alt + Click was only used to mark gap
junctions. With the introduction of more connector types, we started to
use the context menu on Alt + Click when a node is selected. I changed
this behavior so that on Alt + Click now always a menu is shown,
regardless of what node type is selected. This I feel is the least
surprising mode of operation, plus it provides the most flexibility.

>3. Only attachments marked with method 2.a) show up in the connectivity
>widget. Is this intentional?

Yes, because the Connectivity Widget displays connections between
skeletons, which means both the link from skeleton A to a connector and
the link from the connector to skeleton B are looked at. Only a
particular set of link type combinations is supported by the
Connectivity Widget: pre->post, post<-pre, gapjunction-gapjunction,
attachment-close_to. For attachment connectors this currently means that
they are only shown if skeleton A and B are linked by A using the
"attached_to" relation with a connector and skeleton B using the
"close_to" relation to link to that connector (which are automatically
used with Shift + Click on an attachment connector, link arrows are
shown in yellow).This is admittedly not obvious at all, but is the
semantic I chose attachment connectors to have at some point to be
easier to extend.

By the way, you can also show other connection types in the Graph
Widget, by selecting the representative link types from the "Properties"
dialog or the "Edges" tab.

>3. When listing connections in the connectivity widget, attachments
>only show up on the "pre-synaptic" neuron (attachments_2018-10-37.png shows the
>"post-synaptic" neuron). But aren't attachments supposed to be
>non-directional?

Well, not really. Like said above, originally I created them to
represent things that were attached to a particular location in a
skeleton (e.g. vesicles) and not necessarily two neurons. In this
context the neuron with the "attached_to" relation (the "first" relation
created to the connector) is sort of the "origin" of these links. With
dedicated connector types,

>4. Setting attachment as the default connector type doesn't work and I
>get a warning (see unknown_connector_type_2018-10-31.png). This only
>happens with the attachment connector type; synaptic, abutting, and gap
>junction all work.

Thanks for reporting this. Shift + Click with attachment connectors
should work now if they are set as default connector type.

Best,
Tom

[Tom Kazimiers]

Hi Sanja,

On Sat, Nov 03, 2018 at 06:02:28PM -0400, Tom Kazimiers wrote:
>Looking at your screenshot though it looks like you would want to link
>two skeletons through such a junction connector or desmosomes
>connector. In that case I believe it might be better if I would just
>add these connector types to CATMAID itself. I suppose both types would
>work like gap junctions, i.e. non-directional and a maximum of two
>skeletons can be linked with a gap junction connector? This would make
>it easier to look at the two types separately from each other. It would
>be relatively easy to add support in the creation context menu, default
>connector types, Connectivtiy Widget, Graph Widget and 3D Viewer.

I looked into this a bit more and added the two new connector types you
mentioned: tight junctions and desmosomes. They work just like the gap
junction connector, i.e. non-directional and two links per connector
are allowed. Let me know if they should have different properties. The
connectivity widget comes now also with a control that lets you select
all available link types rather than only attachment and gap junction.
The Graph Widget supports the new connector types as well, but the 3D
Viewer and Connectivity Matrix don't support them yet.

Cheers,
Tom

[Sanja Jasek]

Hi Tom,

Thanks for the reply, and sorry it took me so long to reply now.

Looks like I completely misunderstood the connector types. Thanks for the clarification.

Thanks so much for this. You understood correctly, desmosomes should be similar to gap junction connectors. The only difference is that sometimes they are hemidesmosomes, that is they don't have another skeleton as a partner, but only attach to the basal lamina. One possible solution for desmosomes that are linked to only one skeleton is to default the other partner skeleton (that doesn't exist) to the basal lamina, and list it as such in the connectivity widget.

For the other connector, we are actually not sure what kind of junction they are at the moment. Maybe the connector can just be called "junction"? Or you can leave whatever name you gave it and we can just keep in mind to call it something else when we are confident what exactly it is. In any case, it should be exactly the same as a gap junction. This one is not so important to us at the moment. The desmosome connector is much more important. It would be great if we could select it as the default connector in preferences.

Thanks again for your help.

Best regards,

Sanja

[Tom Kazimiers]

Hi Sanja,

On Wed, Nov 07, 2018 at 02:46:27PM +0000, Sanja Jasek wrote:
>Thanks so much for this. You understood correctly, desmosomes should be
>similar to gap junction connectors. The only difference is that sometimes
>they are hemidesmosomes, that is they don't have another skeleton as a
>partner, but only attach to the basal lamina. One possible solution for
>desmosomes that are linked to only one skeleton is to default the other
>partner skeleton (that doesn't exist) to the basal lamina, and list it as
>such in the connectivity widget.

You are welcome! What you could do to represent hemidesmosomes is to
make the "other partner" just a new skeleton (a single node is enough)
at the basal lamina side of the hemidesmosome. This would be needed to
list those links in the Conenctivity Widget and other places. You can
then either give this new skeleton an annotation (using e.g. the F3 key)
like "Basal lamina" and/or name the skeleton like this (using e.g. the
F2 key). Some widgets allow grouping based on naming patterns, so you
could treat all hemidomes as if they would connect to the same "Basal
lamina" skeleton. The Connectivity Widget unfortunately can't do this at
the moment though, but e.g. the Connectivity Matrix and Graph Widget
can.

>For the other connector, we are actually not sure what kind of junction
>they are at the moment. Maybe the connector can just be called "junction"?
>Or you can leave whatever name you gave it and we can just keep in mind to
>call it something else when we are confident what exactly it is. In any
>case, it should be exactly the same as a gap junction. This one is not so
>important to us at the moment.

I see. For now I added the desmosome and the tight junction link type.
Maybe it would make sense to have a generic junction connector for
situations like yours. I doubt though that I will find time to add this
before the next release. An alternative for now would be to maybe use
tight junctions connectos and add a tag to the connector like TODO
(using the L key---see help with F1 for more details on tag short
cuts---or use T for custom tag names). Later this would allow you to
quickly list all TODO tagged connectors and either change their type to
another link type or remove the tag if they are tight junctions.

> The desmosome connector is much more important. It would be great if
> we could select it as the default connector in preferences.

Yes, you should be able to select this connector type as default.

Best,
Tom

[Sanja Jasek]

Hi Tom,

On Thu, 8 Nov 2018 at 05:01, Tom Kazimiers <t...@voodoo-arts.net> wrote:

You are welcome! What you could do to represent hemidesmosomes is to
make the "other partner" just a new skeleton (a single node is enough)
at the basal lamina side of the hemidesmosome. This would be needed to
list those links in the Conenctivity Widget and other places. You can
then either give this new skeleton an annotation (using e.g. the F3 key)
like "Basal lamina" and/or name the skeleton like this (using e.g. the
F2 key). Some widgets allow grouping based on naming patterns, so you
could treat all hemidomes as if they would connect to the same "Basal
lamina" skeleton. The Connectivity Widget unfortunately can't do this at
the moment though, but e.g. the Connectivity Matrix and Graph Widget
can.

That's a good idea. I'll do it that way.

 

>For the other connector, we are actually not sure what kind of junction
>they are at the moment. Maybe the connector can just be called "junction"?
>Or you can leave whatever name you gave it and we can just keep in mind to
>call it something else when we are confident what exactly it is. In any
>case, it should be exactly the same as a gap junction. This one is not so
>important to us at the moment.

I see. For now I added the desmosome and the tight junction link type. 
Maybe it would make sense to have a generic junction connector for
situations like yours. I doubt though that I will find time to add this
before the next release. An alternative for now would be to maybe use
tight junctions connectos and add a tag to the connector like TODO
(using the L key---see help with F1 for more details on tag short
cuts---or use T for custom tag names). Later this would allow you to
quickly list all TODO tagged connectors and either change their type to
another link type or remove the tag if they are tight junctions.

That's fine. Thank you.

 

> The desmosome connector is much more important. It would be great if
> we could select it as the default connector in preferences.

Yes, you should be able to select this connector type as default.

It works. Thank you very much.

Best regards,

Sanja