CHANGELOG.md

2015.12.21

Contributors: Albert Cardona, Andrew Champion, Eric Trautman, Tom Kazimiers

Features and enhancements

Selection table:

• A new option ("Append with batch color") allows to override the color and opacity of appended skeletons with the current batch color and batch opacity. It is deselected by default.

• Clearing the table doesn't ask for confirmation anymore. This has been removed for consistency with other widgets and because of the now available option to save/open skeleton lists.

New widget "Synapse Distribution Plot":

• For one or more neurons, plot distances of postsynaptic sites relative to an axon initial segment, represented by a skeleton node that is either computed or given via text tags. Each row represents the inputs contributed by a presynaptic arbor. Presynaptic arbors (rows) are sorted from more to less synapses. Presynaptic arbors can be filtered (i.e. hidden away) by a threshold on the number of inputs each provides, or by being listed in another widget that has selected skeletons. Individual synapses take by default the color of the postsynaptic arbor (that is, the arbors added via "Append"), but can be colored as well according to neuron colors in another widget. Click on an individual postsynaptic site to go to the corresponding skeleton node. Click on the legend to jump to the skeleton node representing the axon initial segment, relative to which all distance measurements where made. All presynaptic neurons are available in a separate skeleton source for each widget to pull neurons from.

New widget "Synapse Fraction":

• For one or more neurons, render a normalized stacked bar chart with the number of synapses for/from each partner skeletons (directly either upstream or downstream). Can group partner skeletons: add groups by selecting a skeleton source (another widget listing skeletons). Click on the legend to edit (title, color) the group, or remove it. Click on the legend to go to the nearest node in the partner skeleton. To open this new widget, open a connectivity widget and push the button named "Open partner chart".

Settings widget:

• Persistent settings are now scoped, so that default settings may be configured for an entire CATMAID instance ("global"), for each project ("project"), for each user ("user"), and for each user for each project ("session"). Only administrators have access to change project settings, and only superusers have access to change global settings. This allows, for example, administrators to set up recommended defaults for projects so that users only need to adjust their settings where their preferences differ from the project defaults. A selection box for which scope to adjust is at the top of the settings widget. Persistent settings will display information about scopes and defaults when hovering over them with the cursor.

• Administrators may also lock persistent settings so that global or project defaults can not be changed by users.

Graph widget:

• In the main tab, a new remove button removes skeletons in the selected skeleton source from the graph.

3D Viewer:

• Added new buttons under the "Export" tab to export connectors and synapses as CSV. And Skeletons are now exported to CSV with a new column, the radius at each skeleton node, and another new column for the neuron name as rendered by the NeuronNameService (controlled by the Settings). The connectors CSV contains, for each row, the connector ID, the treenode ID, the skeleton ID and the relation ID, mimicking the treenode_connector table. The synapses CSV exports two files: 1. The skeleton ID vs the neuron name 2. The list of synaptic relations of any arbor visible in the 3D Viewer, with columns for the presynaptic skeleton ID, its treenode ID emitting the synapse, the postsynaptic skeleton ID, and its treenode ID that receives the synapse.

Review system:

• Creation, deletion, and edition of synapses and relations now causes related nodes to become unreviewed. Changes to presynaptic relations or the connector itself cause all related treenodes to become unreviewed, while changes to postsynaptic relations affect only that specific related treenode. Changes to other connection relations (abutment, etc.) behave like presynaptic relations, propagating to all related treenodes.

Skeleton source subscriptions:

• So far some widgets allowed to synchronize their skeleton list along with individual property changes. This was done through a "Sync to" selection which pushed this information to other widgets. This has now been replaced with a subscription option. Many widgets allow now to react to changes in skeleton lists in other widgets. Widgets supporting this got a new small chain icon in their title bar with which a subscription management user interface can be shown and hidden. Widgets that contain multiple sources, like the connectivity matrix, have one icon per source. A hover title will show which one to use for each source.

• The UI allows to add subscriptions to multiple sources which can then be combined through set operations. Currently sources are combined in a strict left-associative fashion from top to bottom of the list. When "Override existing" is checked, widget local skeletons are not used when subscriptions are refreshed and will subsequently be removed. Otherwise, the local set is united with the first subscription before all other subscription sources are applied.

• The scope of individual subscriptions can be adjusted: By default each subscription reacts to skeletons added, removed and updated in a source. The "Filter" selection allows to listen to only one of these events. For instance, subscribing to the active skeleton with the "Only additions" filter, allows to collect skeletons selected active skeletons without removing them again from a widget.

• By default, only selected skeletons are subscribed to. This means if a skeleton becomes unselected in a source it is removed from the target widget. If the "Only selected" checkbox is unchecked, also unselected skeletons are added to a target widget. They are removed when skeletons are removed from the source and their selection state is synced.

• All widget still feature the "From [Source] Append/Clear/Refresh" work-flow. The subscription UI's "Pull" button does the same as the regular "Append" button: a one-time sync from a source.

Miscellaneous:

• Many tracing widgets now allow a user to hide their controls. A little gear icon in their title bar toggles their visibility.

• Rather than only specifying annotations that are used as successive fallbacks for labeling neurons, neuron labels can now be specified as arbitrary combinations of annotation-based components using a format string. This is still configured in the annotations section of the settings widget.

• Volumes can now be edited when clicked on in the volume widget. This will also display the edited volume as layer in the active stack viewer.

• Moving a node in the tracing overlay now updates its position in the database as soon as the mouse is released, rather than waiting until the section changes.

• Changes to the CATMAID API are now documented in API_CHANGELOG.md.

• A docker image of a running CATMAID instance is now available for evaluating or developing CATMAID without needing to perform a complete install. The latest release is available via the "stable" tag, and the current development version is available via the "latest" tag. To try it:

  docker run -p 8080:80 aschampion/catmaid
  

  Then point your browser to http://localhost:8080. The default superuser has username "admin" and password "admin".

Bug fixes

• API testing URL generated by Swagger (used for the API documentation at /apis) now respect a sub-directory that CATMAID might run from.

• Fixed near active node shading in the 3D viewer.

• Pre- and post-synaptic edges in the tracing overlay now update when dragging a related treenode.

2015.11.16

Contributors: Albert Cardona, Andrew Champion, Tom Kazimiers

Features and enhancements

Key shortcuts / mouse operations:

• Pressing \ now brings up a dialog to go to the nearest node with a label matching a query regex. If a node is active, this search is limited to the active skeleton. Shift + \ cycles through matching nodes in ascending distance order. Ctrl + \ repeats the last search regex without prompting.

• Deleting a virtual node with DELETE or Ctrl+Shift+click instead suppresses the virtual node. Suppressed virtual nodes are skipped during review. A setting is available to also skip suppressed virtual nodes during normal navigation with [ and ].

Selection table:

• Batch coloring is now much quicker.

• If the batch color button is pressed a second time the color picker will not only close but also will the batch coloring be re-applied. This won't happen if the color picker is closed by clicking somewhere else.

• The status text line at the bottom of the table includes now the number of selected neurons. This is helpful when a filter is active and more neurons are selected than visible.

• Sorting for visibility columns has been removed.

• Neurons part of a Selection Table can now also be filtered based on annotations. This can be done with the help of the input field next to the name filter. Like with the name filter input, pressing the Enter key will activate both filters and starting with a slash character ("/") will make the input be treated as a regular expression (to e.g. only show neurons that are annotated with a1 and b2, use "/a1|b2"). For now no meta-annotations are taken into account.

• With the help of the new "Open" and "Close" buttons, skeleton lists can be stored into JSON files as well as loaded from them. Along with each skeleton ID, the current color and opacity is stored.

Skeleton projection layer

• With the new "Skeleton color gradient" coloring mode, the skeleton's tracing color (currently only yellow fo the active skeleton) is used for coloring. It fades into downstream and upstream colors, respectively (which are black and white by default).

• Nodes can be selected by placing the mouse close to them (regardless if displayed or not) and pressing 'g'. If no node is found in close proximity (<50px screen space), the tracing layer's node selection is used.

Graph widget:

• Synapses can be filtered from edges based on their confidence. The confidence threshold is applied to the minimum of the pre- and post-synaptic relation confidences. Confidence filtering is applied prior to synapse count filtering.

• Synapse count coloring on edges can now be configured independently from edge colors.

Volumes:

• A new widget, the Volume Manager, allows to create and list volumetric geometries. These geometries are not yet displayable and for now only box volumes can be created. The widget is available through a new 3D box icon, last in the list of tracing tool widgets.

• New nodes can now be tested for intersection with a certain volume. The Tracing section of the settings widget allows to choose a volume against which new nodes will be tested. If they are outside of it, a warning will be shown.

Neuron Search:

• Partial annotations as well as regular expressions are now supported for searching. If the text entered in an 'annotated' search field matches a single existing annotation (i.e. one that would also show up in the auto-completion), it is used as search constraint, just like before. However, if no matching annotation was found, the input text is treated as a regular expression on annotation names if it starts with a slash character ('/'), otherwise it is treated as a regular search pattern over all annotations. For instance, finding all things that are are annotated by either A1 or B2 would look like '/A1|B2' or requiring annotations that end on 'xyz' could be searched for by '/xyz$'. This also works with sub-annotation queries.

3D viewer:

• Different neuron visibility modes are now available for animations. A drop down list replaces the check-box and an option dialog is shown if a particular animation mode requires user input. Besides the 'show one neuron per rotation' mode, there is now also the 'Show n neurons per rotation' mode and a mode which uses a pattern to explicitly define the visibility of particular neurons after a particular rotation. The animation export now uses the visibility mode selected in the 3D viewer.

Administration:

• CATMAID has been able to use DVID as a project/stack back-end and as a image source for quite a wile now. To make the latter option easier to setup, a new admin tool is available to create CATMAID stacks based on a DVID server. It can be found in the "Custom views" section of CATMAID's admin interface, labeled as "DVID stack importer". With the help of this tool on can inspect all available repositories and data instances on a DVID server and create a stack based on one data instance. CATMAID will make sure that all pre-conditions are met by a stack created this way.

Miscellaneous:

• By default new widgets will now select the last widget created as skeleton source. If wanted, this can be adjusted to the previous behavior (always select 'Active skeleton') through the 'Auto-select widget created last as source for new widgets' option in the settings widget.

• Multiple stacks opened through a URL can now optionally be opened in the same stack viewer window by adding "&composite=1" to the URL.

• If an already reviewed node is moved it will now become unreviewed again.

• Links clicked in the message menu will now open in a new page.

Bug fixes

• The skeleton projection layer will now update automatically on skeleton changes like new or removed nodes as well as splits and merges. It will also not complain anymore if a connector was selected.

• Text rendered in the 3D viewer is now upright again (instead of upside-down).

2015.10.19

Contributors: Albert Cardona, Andrew Champion, Tom Kazimiers

Features and enhancements

Scripting:

• The widget instance associated with the focused window can be retrieved with the convenience function CATMAID.front().

Orthogonal views and multi-channel data:

• Stack groups can be used to relate different stacks to each other within one project, e.g. to make clear that some stacks are different orthogonal views or different channels of the same dataset. If there are stack groups defined in a project, they are for now available through the "Projects" menu, which provides sub-menus for stacks and stack groups for each project. When opened, the stacks of a channel based stack groups are added as layers to the first stack. Ortho-view stacks are all opened in a separate stack viewer.

• If a stack group consists of the three different orthogonal views for a dataset, the window layout is adapted automatically as soon as the stack group is opened. The layout will be a four-pane layout in which the left half of the screen is the XY view on top of the XZ view and the right half of the screen is the ZY view on top of a selection table.

• Since stack group are instances of the 'stackgroup' class, they can be referenced from within ontologies. All projects now have a 'stackgroup' class and the relations 'has_view' and 'has_channel' created by default. They are also created for projects that don't have them, yet.

• Stack groups can be created and managed from with the admin interface through either the new Stack group page or while editing/creating a stack.

3D viewer:

• Skeletons can be shaded by distance from a plane through the active node. The plane can either be a Z-plane in project space or a plane normal to the ray from the camera to the active node.

• New "Count" button to count the number of pre- or postsynaptic sites, or the number of treenodes tagged with a given text tag, within a distance of the active node in the selected arbor along the cable of the arbor, or within a given Euclidean distance for any arbor present in the 3D viewer.

Tile layer:

• WebGL rendering is now compatible with orthogonal views.

• Tiles can now be rendered either with linear pixel interpolation (previous default behavior) or nearest neighbor interpolation. This is controlled by the "Image tile interpolation" setting.

Graph widget:

• When growing by circles, the set of neurons added can be filtered to include only those with annotations matching a regex.

Miscellaneous:

• A new color picker replaces the color wheel. The new control hovers over other elements so it can be moved, has color memory slots, defaults to a smaller size and can be resized to show input elements for different color spaces. To save a color in a memory slot, click on the rectangle containing the small circle next to the memory slots.

• Documentation for some HTTP API endpoints is now available from your CATMAID server at the /apis/ URL.

Bug fixes

Tile layer:

• Fixed a major WebGL tile layer GPU memory leak.

3D viewer:

• In orthographic mode, the correct depth ordering is now used again.

Selection table:

• Color sorting works again.

Miscellaneous:

• An error no longer appears when selecting an un-annotated skeleton while neuron labels are configured to use a meta-annotation.

2015.9.11

Contributors: Albert Cardona, Andrew Champion, Tom Kazimiers

Features and enhancements

Neuron Navigator:

• It is now possible to remove multiple annotations at once from a neuron. A new column is added to annotation tables, each annotation row has now a checkbox in its first column. A click on the de-annotate link in this column's header or footer will remove all selected annotations from the current neuron.

Tracing:

• If a single-node skeleton is merged into another skeleton, no merge dialog is now shown by default. All annotations of this single-node skeleton are merged into the target skeleton without asking. This behavior can be changed to again show a merge UI if the single-node skeleton has annotations (behavior so far) through a new entry in the Tracing section of the settings widget.

Graph widget:

• New layout modes "dagre", "cola", "spread" and "springy". The first is based on DAGs (directed acyclic graphs) and the last three are force-directed. To note that "spread" will evenly layout neurons in trying to occupy as much space as possible, and also leads to symmetric-looking graphs when rendering multiple disconnected graphs of e.g. left and right homologous neurons. Try it.

Connectivity matrix:

• Descending and ascending sorting is now available for all sorting modes.

• The new 'order of other' sorting mode will try to follow the column order for rows and vice versa. If skeletons are not found in the reference dimension, they are pushed to the end.

Selection table:

• When changing the color of a neuron, all other selected neurons can be colored at the same time, when the new 'all selected' checkbox (right above the color wheel) is checked.

Neuron sarch:

• The "select all" checkbox now does what it says and selects all neurons in the results set.

• Pagination now works like in other widgets and the number of elements per page can be adjusted.

• Annotations are not loaded by default anymore, but can be shown with the help of the new "Show annotations" checkbox.

Miscellaneous:

• When a connector is selected, basic information about it is displayed at the top of the window (where otherwise the neuron name is displayed).

• A neuron search result's annotation list is now kept in sync with the rest of the widgets. If annotations change on a listed neuron or annotation, the search query is re-done.

• If only transparency is changed in the batch color picker of the Selection Table, the transparency alone (and not the color) of the target skeletons will be updated. To also update the skeleton color, the color has to be changed in the color picker, too.

Administration:

• Adding custom code to CATMAID's front end is now easier: Add file names to the STATIC_EXTENSION_FILES array variable and have your web-server point the URL defined in STATIC_EXTENSION_URL (defaults to /staticext/) to the folder were those files live. CATMAID will then load those files after its own files.

Bug fixes

• Nodes are now correctly scaled in skeleton projection layers.

• Neuron navigator now updates if a skeleton is changed (e.g. due to a merge).

• 'Sync to' selections to push changes from one widget to another (e.g. 3D viewer controlled by selection table) are now updated correctly, if a selected target is closed.

• Changing the order of rows and columns of the connectivity matrix manually does now work like expected.

• From within the neuron search removed annotations will now disappear again from the search widget after they are unlinked.

• Using the CATMAID coloring scheme in the Selection Table is not random anymore.

• CSV files exported from the Connectivity Widget now respect the table ordering and include the target neuron names.

• Spheres that intersect in the 3D viewer (e.g. somas) don't appear broken up anymore.

• The 3D viewer's SVG export will now correctly calculate the size of exported spheres (e.g. soma tags).

2015.7.31

Contributors: Albert Cardona, Andrew Champion, Tom Kazimiers

Features and enhancements

Key shortcuts / mouse operations:

• Ctrl + [ or ] now navigates to the next real (non-virtual) parent or child of the active node, respectively.

Connectivity widget:

• Upstream and downstream partners can now be filtered by synaptic confidence in addition to synaptic count. Synaptic confidence filtering is applied before count filtering. Confidence is taken to be the minimum of the presynaptic and postsynaptic connector confidence.

3D Viewer:

• Tags matching a custom regex can be shown as handle spheres in addition to tags containing "uncertain" or "end". Note that after entering the regex in the "View Settings" tab the viewer must be refreshed from the "Main" tab before tag spheres are updated.

• The active node marker now resizes based on the radius of the active node.

• The active node marker can now optionally always be drawn on top of other objects, even if it is occluded.

Miscellaneous:

• A default neuron name can now be specified in the settings widget. Similar to annotations, the pattern "{nX}" can be used to add an automatically incrementing number to each new neuron created, starting at X. Omitting X will be interpreted to start from 1. This default name does not persist between sessions.

• Neuron navigator: neuron name and annotation search are now case-insensitive.

• The client now checks every 15 minutes whether it is the same version as the server. If not, an error dialog is shown prompting the user to refresh their browser.

• It is now possible to show a projection of the active skeleton in the tracing overlay. All nodes will be displayed in the current slice, but no interaction is allowed with them. This feature can be useful to get more context on the current location in the active skeleton. This mode can be toggled with F10 or through a new entry in the settings (Tracing Overlay > Active skeleton projection), where different parameters can be adjusted. The color for the upstream and downstream part can be independently changed and various shading modes can be selected (plain color, Z distance transparency, Strahler based transparency or cut off).

Bug fixes

• 3D viewer: the correct synapse colors are now used when connectors are restricted.

• If annotations are added or removed, annotation search widgets are updated correctly. You can now search for newly created annotations without having to open a new search widget.

• The order of the selection table is now remembered before it is refreshed.

• Connectivity widget: common partners filtering now correctly enforces that partners are partnered with all target neurons, not just any two.

• Review widget: the skipped node warning will now only show up when in fact more nodes have been skipped than allowed.

• Fixed a vulnerability that allowed any user with "browse" access permissions to any project to execute arbitrary SQL statements on the CATMAID database.

2015.7.17

Contributors: Albert Cardona, Andrew Champion, Tom Kazimiers

Features and enhancements

Connectivity widget:

• Partners can now be filtered by a minimum threshold on number of nodes, rather than only being able to filter single-node partners.

• The user's review team is now an option in the partner review filter.

• Changing the review filter no longer reloads the entire widget.

• Many small performance improvements.

Selection table:

• The table layout has been streamlined with other tables in CATMAID. All columns except 'action columns' can now be used for sorting. Pagination is now done with the buttons to the right above and below the table, the page length can now be adjusted, too. The summary info button moved into the button panel while the filter input is now part of the table.

• It is now possible to add a new annotation to individual neurons without changing the current selection. This can be done with the little tag icon in the actions column on the right. The former info button was replaced by a small 'i' icon and clicking the folder icon in the same column will open a Neuron Navigator window for the respective neuron.

• All visibility related colums can be hidden with a new checkbox in the button panel. This might be useful to save space if a selection table is not used to control a 3D viewer.

Miscellaneous:

• Neuron search: annotations can now be searched for those by users in the review team.

• Log: entries can now be filtered to include only actions from the user's review team.

• The maximum number of nodes returned to the tracing overlay is now configurable as a server setting: NODE_LIST_MAXIMUM_COUNT (default 5000).

• Group graph: a new lock buttons in the Selection tab allows to lock selected nodes, so that their position doesn't change until they are unlocked again.

Bug fixes

• Fix display of intermediate nodes on edge between two joined skeletons.

• The last lines of the review widget were hidden sometimes. This is not the case anymore.

2015.7.6

CATMAID now uses the GPLv3 license and moved away from the stricter AGPLv3. This move was discussed with all previous contributors and agreed on. See the corresponding commit for more details.

Contributors: Albert Cardona, Andrew Champion, Tom Kazimiers

Notes

This release includes database changes that require manual intervention if you are upgrading from an existing installation. A new dependency is now required: PostGIS, an extension to PostgreSQL. After its installation, it has to be activated for the CATMAID database. To do so, connect to the database using a Postgres system user. Assuming a Postgres system user named "postgres" and a CATMAID database named "catmaid", this could be done by calling

sudo -u postgres psql -d catmaid

Being connected to the database, PostGIS can be enabled by executing

CREATE EXTENSION postgis;

Now PostGIS is enabled and the connection can be closed again. Now a regular update can be performed. Please note that this update can take quite some time to complete. On bigger neuron tracing installations, multiple hours are realistic.

Features and enhancements

Key shortcuts / mouse operations:

• Ctrl + mouse wheel now zooms the stack, while shift zooms by smaller increments.

• Pressing X in the tracing tool will begin measuring distance from the current cursor position. The mouse wheel can be used to measure along the stack Z axis. Clicking will close the measurement tool and show the final distance in the status bar. ESC cancels the tool.

Multi-view tracing and virtual nodes:

• Orthogonal views on a regular XY stack can now also be used for neuron reconstruction. If they are available as CATMAID stacks and opened while the tracing tool is activated, tracing data will be shown in the respective orthogonal views as well. Tracing can be done in these views just like in the regular XY view.

• When tracing, it is not required anymore to place a node in every section. If no node has been placed in a section, CATMAID will place a so called virtual node where the skeleton and the section meet. If this virtual node is modified in any way, e.g. tagging, joining, moving, etc. it will be created. This also slightly changes the way reviews work. Review information is only stored on real nodes.

• The review widget has a new settings: in-between node step. It specifies how many sections can be skipped between adjacent real nodes. This is done with respect to the currently focused stack. This stack is also used to determine in which direction to move to look beyond the start of a segment.

Stack viewer:

• Other stacks in a project can be added to an open stack view by selecting the "Add to focused viewer" option from the stacks menu. This allows multiple stacks to exist in the same view like overlays, while accounting for differences in translation and resolution. The navigator will expand the available zoom levels to accomodate the maximum and minimum zoom possible in all of the open stacks.

• Tile layers for stacks added to a viewer can be removed from a viewer via an "x" in the tile layer control.

• Multiple viewers into the same stack can now be opened.

• Each stack viewer can be toggled between coupling its navigation with other open stack viewers. Toggle this via the "Navigate with project" checkbox in the tile layer control.

Tile layer:

• New filter (WebGL only): intensity thresholded transparency

3D Viewer:

• Connector restriction can now explicitly be turned on and off with a pull down list. One can select between "Show all" (i.e. restriction turned off), "All shared connectors" will only show connectors with partners in the current selection and "All pre->post connectors" will only allow connectors with at least one presynaptic partner and one postsynaptic partner in the current selection. The last option, "All group shared" allows to select two skeleton sources (e.g. two selection table) and it will only show connectors that are part of the 3D viewer and that connect between both selected groups of skeletons. There is also a pre->post enforcing variant of it.

Neuron search:

• The name displayed for neurons now follows the same naming mechanism used for other widgets. It can be controlled through the settings widget and will automatically update if a neuron name is changed.

Miscellaneous:

• In the connectivity widget, upstream and downstream thresholds can now be set at once for all seed neurons. Two drop down controls used for this will be displayed if there is more than one seed neuron.

• Treenode Table refurbishing: far faster, supports multiple skeletons, can do tag search with regular expressions and lists the skeleton treenode ID.

• Rows and columns of the connectivity matrix can now be moved around with little buttons that appear when the mouse is over row and column header cells.

• There are now three options to change focus if a pointer enters a window: don't change focus (how it has been so far), focus stacks (will activate stacks when hovered, but won't change focus for other windows) and focus all (will change focus to every window hovered). The default will be stack focus follows the pointer. The settings widget makes these options available in general settings area.

• There are three new controls in the split/merge dialog: Toggle the display of input and output markers on neurons in the embedded 3D viewer and select which shading method is used (default: "active node split"). Alternatively, "Strahler index" coloring can be used, which helps with depth perception.

• Attempting to reload a CATMAID browser tab or go back in history, will now result in a warning dialog, asking for confirmation. It makes clear that CATMAID's window layout and content won't be saved if the acton isn't canceled.

Administration:

• Now that virtual nodes are available, existing database can (but don't have to) be optimized. A new management command will look for straight skeleton parts that are not referenced in any way and prunes them. In other words, if there are three successive collinear nodes and the middle one is not referenced, it will be removed.

  manage.py catmaid_prune_skeletons

Bug fixes

• The Neuron Search widget doesn't throw an error anymore when a neuron listed in it is merged.

• There is no longer a race condition in the database during concurrent split/merge of a skeleton and creation of a treenode in that skeleton. While this is not a comprehensive guarantee of conflict-free concurrency, it does remove the most likely scenario resulting in corruption of the database model.

2015.5.27

Bug fixes

• Fix radius based neuron selection in tracing window.

2015.5.19

Contributors: Tom Kazimiers

Features and enhancements

Key shortcuts / mouse operations:

• Cycling through open end nodes will now only visit the root node if it is an actual leaf. That is, when it has only one child node and is untagged.

3D Viewer:

• A light background shading variant for connectors was added. It uses a darker cyan color which provides more contrast if a white background is used.

Miscellaneous:

• The location of messages and notifications can be configured in the settings widget. The default location is still the upper right corner.

• If the node display limit is hit while panning the field of view in tracing mode, node refresh will be temporary disabled. Once the mouse button is released again an no further panning happens within one second, node update is reset to normal. This allows for smoother panning if many nodes are visible.

Bug fixes

Review system:

• Review teams are now respected when Shift + W is used to jump to the next unreviewed node.

3D viewer:

• Skeletons with other coloring than "Source", will now be visible when exported as SVG in the 3D viewer.

Miscellaneous:

• Skeletons added to a selection table, will now honor the table's "global" settings for pre, post, meta and text visibility.

• If an annotation is removed from a neuron, the annotation itself will be deleted, too, if it is not used anywhere else. Now also meta annotations of the deleted annotation will be removed (and their meta annotations...), if they are not used anywhere else.

2015.5.11

Contributors: Albert Cardona, Andrew Champion, Tom Kazimiers

Features and enhancements

Connectivity widget:

• Partner filtering now supports regular expressions when the first character of the search input is "/".

Treenode table:

• REMOVED: Radii can no longer be edited by clicking on their cell in the table.

Connectivity matrix:

• The tracing tool has got a new widget: a connectivity matrix. It can be opened with the "M" frame icon next to the button for the connectivity widget. To use it, one has to append skeletons for its rows and columns. Skeletons can also be added as group. Each cell shows two sub-cells, the first one shows the number of synapses from row to column and the second one the number synapses from column to row. When a synapse count number is clicked, a connector selection is opened, that contains the corresponding synapses. Both pre- and post-synaptic count cells can be colored individually. By default a coloring similar to the tracing layer's red and cyan is used. There are also color gradients available to produce heat maps (i.e. color cells based on the actual synapse count).

• Graph widget: ability to split neurons by text tag on their skeletons. It's the "Tag" button under the "Subgraph" tab. Enables you to manually define regions on a neuronal arbor (like axon and dendrite, or multiple dendritic domains) and then have them be represented each as a node in the graph. The skeleton will be virtually cut at the nodes containing the tags, with the tagged node belonging to the downstream part (relative to the root node).

3D Viewer:

• Color mode "Downstream of tag" is now a shading mode.

• New synapse coloring mode "Same as skeleton". If you then hide the skeletons and show only the synapses you will e.g. see spatial tiling of ORN axons, each defining a glomerulus in the larval olfactory lobe.

• For PNG and SVG export one can now specify the dimensions of the result files. A dialog shown before exporting asks for width and height.

Neuron dendrogram:

• The horizontal and vertical spacing between nodes in the neuron dendrogram can now be fine tuned.

Administration:

• A new tool 'Group membership helper' has been added to add multiple users to multiple groups or to revoke their group membership. This can be used to control access over the data created by individual users.

Miscellaneous:

• A node-placement-and-radius-edit mode has been added. If enabled through the settings widget (Tracing > "Edit radius after node creation"), the radius for a node will be edited immediately after it has been created. This allows for easier volumetric reconstruction. In this mode, the radius circle editing tool is used to specify the radius. No dialog is shown once a radius is selected for a node and it will only be saved for the new node.

• A new connector type ("abutting") can now be created. In contrast to the regular synaptic connector, it can be used to represent the fact that two or more neurons are in abutting processes. For now this mode can be activated through the settings widget (Tracing > "Create abutting connectors"). For abutting connectors the lines representing the links to nodes will appear in a green color.

Bug fixes

3D viewer:

• Adding and removing neurons and static data lead in some situations to many errors that were displayed on the console (and therefore not visible to most users) and caused minor performance problems. This has been fixed and all data should now be added and removed correctly.

• Following the active node should now work much more reliable. Before, it could happen that this stopped working after a second 3D viewer was closed.

Connectivity widget:

• Fix one cause of sluggish behavior for widgets that have been modified many times. Also fixes repeated alert dialogs when clicking a neuron in the partner tables that no longer exists or does not have any treenodes.

Neuron Navigator:

• Don't show an error if an invalid regular expression was entered for searching. Instead, color the search box red and show a warning message.

2015.3.31

Contributors: Albert Cardona, Andrew Champion, Tom Kazimiers, Stephan Gerhard

Features and enhancements

Key shortcuts / mouse operations:

• Shift+T removes all tags from the currently active node.

• After using R to go the nearest open leaf, shift+R cycles through other open leaves in the skeleton in order of ascending distance from the starting location. Combining alt with these operations orders open leaves by most recent creation instead of distance.

• Ctrl+Y removes the active skeleton from the last used selection widget.

• Shift+Y selects skeletons within a radius of the active node in the tracing layer and adds them to the last used selection widget. Ctrl+shift+Y works in the same way to remove skeletons from the last used selection widget.

• If the next (or previous) branch/end point is already selected when V (or B) is pressed, the view will center on it nevertheless.

• If the mouse is over the stack when zooming, the view will recenter so that the same stack location remains under the mouse at the new scale.

• During review, Q and W during will refocus on the last reviewed neuron if review is interrupted (another node is selected), regardless of the auto centering setting. If one looks beyond the current segment, the last reviewed node will be selected by Q and W as well, but auto centering is respected.

Review system:

• New "Reviewer Team" system allows filtering reviews in visualizations and statistics to include only those by particular reviewers. Each user can control which reviewers to include in her team. A date can be configured for each reviewer in the team, so that only reviews from that reviewer after this date are included.
  • A user's reviewer team is configured through the Settings widget.
  • The review widget includes a team column between the user and union columns.
  • The percent reviewed column in the selection widget can be set to team or union.
  • Team review coloring is available in the 3D viewer and group graph.

3D viewer:

• With Ctrl + mouse wheel, only the camera is moved in target direction, the target stays fixed. If Alt + mouse wheel is used, the target moves as well.

• The CSV export not also includes the parent ID for each node, which can be used to reconstruct the topological tree.

• The auto-created selection widget is now 50% smaller, giving more vertical space to the 3D viewer.

• With the help of controls of the Animation tab, simple animations can be played. Currently, rotation around the X, Y and Z axis as well as the current "up" direction of the camera. is supported. The back-and-forth mode will reverse rotation direction once a full circle is reached. With the help of the stepwise visibility option, individual neurons can be made visible after a certain amount of time the animation is running. Additionally, neurons can be made sequentially visible after each rotation.

• Animations can also be exported as WebM movie file. The "Export animation" button in the Export tab, will show a dialog with basic export settings. Like with the other view export options, the current 3D view setup is used. The frame size can be adjusted in the export dialog. Creating the file can take some seconds and currently only works for the Chrome browser (due to the lack of WebP support in others). The resulting WebM video file can be converted to any other format using e.g. VLC player, if needed.

• New shading mode "synapse-free chunks". Has one parameter, the minimum amount of synapse-free cable to consider between two consecutive synapses, adjustable from the "Shading Parameters" tab.

• New shading mode "dendritic backbone". Depends on 'microtubules end' tags, or will approximate twigs by using the Strahler number entered in the "Shading Parameters" tab.

• The view settings tab now contains a control to scale the size of the node handles (e.g. active node, special tags).

Tile layer:

• Tiles can now be rendered with WebGL, which enables new visualization features and fixes some flickering issues. Enable via "Prefer WebGL Layers" in Settings. The WebGL renderer is currently considered experimental and may have stability issues on some clients. See catmaid#186 (comment) for details on using WebGL layers with your image stack host.

• The blend mode used to combine stacks and overlays is now configurable when using WebGL. This greatly improves visualization of confocal and other multichannel data. Blend mode is selectable from the layers control, activated via the toggle at the bottom left of the stack view.

• Filters can be applied to layers when using WebGL. Filters can be added and removed from layers through the layers control. Available filters currently include:

  • Gaussian blur
  • Color inversion
  • Brightness, contrast and saturation adjustment
  • Color matrix transform

Connectivity widget:

• It is now possible to remove added neurons again. Each row of the table of target neurons now contains a small 'x' icon in the first column to remove it.

• The selection column is not included anymore in the CSV export.

Analyze Arbor:

• Options are provided to approximate twigs by using a branch Strahler number defined in the "Options".

• Dimensions of the pie charts and XY plots is now configurable from the "Options" dialog.

Graph widget:

• New button to "Clone" the graph widget: opens a new widget with identical content.

• New buttons to "Save" and "Open..." to/from JSON, so that complex graphs can be reloaded later on. Skeletons not present in the database are not loaded.

Miscellaneous:

• Selecting tags for highlighting in the neuron dendrogram

• Synchronization between widgets was improved. Deleting a neuron in one widget, will remove it from other widgets as well.

• Hovering over the CATMAID text on the front page will display CATMAID's version.

Admin:

• For projects, stacks, overlays and data views there is now the option to duplicate objects from within the admin view. To copy objects without their relations, there is now a new action in the list view's action menu. To duplicate an entity with its relations, select the object and use the "save as new" button.

Export:

• A basic JSON export of all treenodes and connectors of the selected neurons is now possible.

Bug fixes

Tracing overlay:

• Trying to remove a non-existent tag from a node now doesn't show an error dialog anymore, but only a notification.

Key shortcuts / mouse operations:

• Fix bug where tagged nodes were not considered open by R regardless of tag content.

Neuron search:

• Make neuron names wrap and use the next line, if there is not enough space for it. This makes the table not expand in width until the name fits anymore.

3D viewer:

• Picking a synapse or other selectable elements is now more robust and now works also in orthographic mode.

• The projection mode (orthographic or perspective) is now also stored in a saved view.

• The 3D viewer's drawing canvas is now correctly sized again. Since the tab panel has been introduced, the 3D viewer has been too high. Now the pre-defined views (XY, XZ, ZY, ZX) are display correctly again, i.e. the whole bounding box is now seen again.

• Performance enhancement when smoothing skeletons with a Gaussian by avoiding to update the same Vector3 instances twice.

Reviews:

• Pressing 'E' during review will now go to the next unreviewed segment as seen from the currently reviewed one. Before, the first unreviewed segment as seen from the top of the table was selected.

• Pressing 'Q' on the first node (leaf) brings one back one section to check if the segment really ends. Pressing 'W' afterwards now brings one back to the first node, not the second like it has been before.

Connectivity widget:

• CSV export works again.

Miscellaneous:

• Vertical resizing of widgets now doesn't lead to strange size changes anymore.

• An alternative DVID tile source was added to support its multiscale API.

2015.1.21

Contributors: Albert Cardona, Andrew Champion, Tom Kazimiers

Features and enhancements

General neuron tracing:

• A new radius editing option has been added that propagates from the current node root-ward to the previous node with an undefined radius (exclusive). Here undefined is taken to be a negative radius, since though the column default is 0 Django initializes it to -1.

Miscellaneous:

• Users need now to confirm the closing of the last stack.

Bug fixes

Tracing overlay:

• A label is now hidden when the mouse hovers over it. Note that this only works for one label at a time, so it is not effective for overlapping labels. A robust solution would require more expensive event propagation over label elements.

• Fullscreen on OS X Safari should now work, too.

• Nodes and arrows are now drawn in order: lines, arrows, nodes, labels

• Fix bug that could occur during radius propagation when the previous node already had a radius defined.

• Fix mouse handlers of node and error drawing, which were broken by adding ordered drawing.

Synapse clustering:

• A long-standing error has been fixed where a few nodes where added to an undefined cluster.

Group graph:

• The root node computation has been fixed.

• Listing edge synapses now also works with split grouped neurons.

3D viewer:

• Make synapse clustering fetch synapses properly (like it is done in the Group Graph).

2015.1.15

Key shortcuts / mouse operations:

• A new shortcut key to navigate to a node's child has been added: ]. It behaves like V by navigating to the largest descendant branch. With Shift+] one cycles through sibling branches in order of descending size.

• For consistency, the P shortcut to navigate to the parent has been replaced with [.

• Navigation to the next branch has changed a bit: The V key now moves to the next branch node or end of the largest descendant branch of the active node, and subsequent presses of shift+V cycle through other possible descending branches in order of decreasing size.

• While editing the radius of a node with the help of the surrounding circle, a click will confirm the current radius (not only pressing 'o' again). The radius editing can now also be canceled with the Esc key.

• With Ctrl+Alt+click one can now insert a node into the active skeleton between two existing nodes.

Zoom:

• Zooming is now also possible in smaller steps. The plus and minus buttons zoom in steps of 1 and with having the Shift key pressed additionally, steps of 0.1 are used.

3D viewer:

• New export options (Export tab):

  • CVS representation of the rendered skeletons;
  • PNG and SVG image of the current view;
  • SVG catalogue of the current view. The catalogue contains each neuron a separate panel on the same SVG document--very useful to generate figures for a paper. Options are provided to sort and arrange panels, and to define pinned neurons that appear in each panel (e.g. a somatosensory axon that acts as reference for each neuron connected to it).

• New "Spatial select" button (Main tab) that allows to select skeletons near the active node or connected to the active skeleton, within a specified distance. Matching skeletons will be shown in a new selection table. This is useful to e.g. select all single-node skeletons connected to the dorsal lobe part of a Kenyon cell.

• Supports orthographic projection (see checkbox in View tab) so that no perspective distortion is applied and distances become comparable between different parts of the view.

• The 3D viewer now has the option to follow the active node (View tab). This acts like clicking "Center active" after each active node change.

• One can bookmark views in the 3D viewer, by pressing "Save view" in the Main tab. Views can be loaded by selecting them from the drop down list next to the button. These bookmarks are currently discarded once CATMAID is reloaded.

• When Ctrl is pressed while zooming in the 3D viewer with the scroll wheel, the camera is actually moved towards its target. This is useful to overcome zooming limits and strong perspective distortion due to a high focal length when zooming.

Selection table:

• "Randomize colors" in the selection table was replaced by a drop down list with different color schemes and the button "Colorize" to apply the selected one. The default is the coloring scheme that existed before. Some of the new color schemes are from Cynthia Brewer (see http://colorbrewer2.org/ ).

• Neurons are activated by clicking on the name, like in all other widgets. The green tick icon has been removed.

• New check box for each neuron called "meta" to toggle the display of extra information like the orange spheres for specially tagged nodes (TODO, uncertain end, etc.) or low confidence nodes.

Dendrogram:

• Can now collapse nodes belonging to a branch that ends in a node tagged "not a branch".

• One can now highlight multiple tags in the dendrogram by separating them with commas.

Graph widget:

• Subgraphs (like axon & dendrite) can now be reset in the graph widget.

Annotations:

• When adding an annotation, the pattern "{nX}" can be used to add an automatically incrementing number to each neuron annotated, starting at X. So if e.g. three neurons are annotated at once with the annotation "test-{n5}", the first one is annotated with "test-5", the second one with "test-6" and the last one with "test-7". Omitting X will be interpreted to start from 1.

• When skeletons are joined, the name of the "losing" skeleton can now be added as an annotation to the "winning" skeleton right in the dialog. Its checkbox is unchecked by default, if the name follows the auto-generated name pattern "neuron 12345".

Searching:

• The neuron name input boxes in both search widgets will now remember entries that have been used before.

Handling the unexpected:

• A general error handler has been added so that CATMAID should hopefully not crash anymore, even if an error occurs. In such situations an error dialog is shown and the error is logged on the server so that we can investigate better what went wrong.

General neuron tracing:

• A robust synapse clustering method was added: centrifugal synapse flow centrality. Many widgets now support a new method for finding axons based on it (e.g. in the 3D viewer as a shading method.

• The connector table now displays the confidence of each link

• Basic import/export support was added. There are two new management commands that can be used by admins to import and export tracing data.

Users and groups:

• Support user registration (disabled by default). Default user groups for new users can be set.

Miscellaneous:

• A new ROI tool was added, which can be activated for each user through the user settings. It currently supports only the creation of new ROIs. Additional sub-tools will be added for more functionality.

Contributors:

This update brought to you by Tom Kazimiers, Andrew Champion, Stephan Gerhard and Albert Cardona.