Merge pull request #6573 from RosettaCommons/roccomoretti/ligand_filter
[Foldit] Update Objectives & Compound Library Panel
One feature that was requested was the ability to filter Compound Library compounds by their matching of objectives. This PR adds a checkbox to the Load Library Compound panel which allows you to do this.
To enable this functionality, I needed to rework all the ligand-based filters. Specifically, I made a new base class from which all of the Filters which act on ligand identities (versus location in the pose) now derive. This allows me to iterate through the current filters and specifically test the RDKit molecules in the compound list. I took the opportunity to do an extensive refactor of how those Objectives were constructed & worked to enable this.
This also reworked/centralized the residue identification parsing code (allowing easy updates later) and also added a "multi" setting to the filter, to possibly change how the objective operates over multiple ligands. Given that all puzzles to date have had a single ligand, I changed the defaults to what I think will be the best approach going forward with the anticipated future ability to split ligands: the filters will sum the settings across all ligands before providing thresholds. (Note that this is possibly at odds with the filtering, which works at a single ligand level. There may be adjustments needed later. I tried to make the interface amenable to that change)
I also added the parent query molecule to the list of molecules presented. This requires support from crowdgames/foldit-server-rails#676 to work, though. (The code is robust to it not being present.)
RELEASE NOTES: Added checkbox to filter Compound Library compounds by (some) Objectives
RELEASE NOTES: Added 'Query' compound to the Compound Library compound list.
Merge pull request #6564 from RosettaCommons/GreenTomato5/interactive_clear_params_cache
[Foldit] Clears Params File Cache for Puzzles Loaded With Secret Buttons
This PR clears the cached params file for puzzles loaded with secret buttons. Currently, puzzles loaded with secret buttons have the same ID of 1, so when a puzzle is first loaded with secret buttons ChemicalManager will use its params file and associate it with a specific tag based on puzzle ID. Because successive puzzles loaded with secret buttons also have the same ID of 1, new params files don't replace the old ones.
Merge pull request #6560 from RosettaCommons/GreenTomato5/interactive_camera_config_on_puzzle_init
[Foldit] Configuration for Camera Starting Position in non-campaign/intro Puzzles
This pr adds support for configuring the starting position of the camera in non intro/campaign puzzles. To configure the starting position the following should be added anywhere in .puzzle_setup fiies:
```
"cam_center" : "x y z"
"cam_eye" : "x y z"
"cam_up" : "x y z"
```
Where x, y, and z for each component are replaced by a position an individual making a puzzle can obtain by using the CTRL+SHIFT+ALT+V keybind and checking their log file. Not adding in these fields to a puzzle_setup file will produce the previous behavior where the camera looks at/goes to a constant position.
Merge pull request #6567 from RosettaCommons/roccomoretti/load_save_tooltips
[Foldit] Make working with saved solutions descriptions easier.
Right now the interface for dealing with saved solutions with descriptions isn't great. There isn't a convenient way of viewing larger descriptions (the display in the LoadSaveDialog cuts it off at 5 lines), and it's not all that convenient to enter the description in the first place.
This PR adds a tooltip in the LoadSaveDialog which displays the full description (inartfully wrapped, if necessary).
It also adjusts the keyboard behavior of the Save Solution panel. Right now, if you hit enter on either the Name or Description, it immediately saves the solution. This disincentivizes adding a description or a longer (multi-line) description. (Which may be why practically no save-with-scientists have a description entry.) Change that such that an enter on the Name moves to the Description field, and an enter on the Description simply adds a line. -- This may or may not be popular with some, as it does slow down the process of saving a bit.
RELEASE NOTES: Show full solution descriptions in Open/Share Solutions panel.
RELEASE NOTES: Adjust how enter and tab keys behave on Save Solutions panel.
Merge pull request #6569 from RosettaCommons/roccomoretti/fix_compound_library_error
[Foldit] Fix issue with Compound Library Objective
It looks like the current stereochemical enumeration we're using with the compound library doesn't work for all molecules. This means that the annotated stored library compound smiles doesn't match what the (post-3D embedding) compound smiles is like.
In particular, a shared-with-scientist molecule from Puzzle 2602 was showing the issue. This particular issue has been fixed in upstream RDKit python code, we can just crib the changes.
Additionally, I've cleaned up a bunch of the code related to CompoundDataStore, which hopefully should make sure that the related functionality (particularly the Compound Library Objective) are more consistent.
RELEASE NOTES: Fix issue where certain compound library compounds weren't getting the bonus.
Merge pull request #6570 from RosettaCommons/roccomoretti/fix_ThreadedRDKitRotamerLibrarySpecification_crash
[Foldit] Fix crash reported for some entries.
Crashes were reported which looked to trace to ThreadedRDKitRotamerLibrarySpecification and its embedding. Make that process more robust. (If we can't get a conformation for the ligands, just skip it.)
RELEASE NOTES: Fix crashes in ligand puzzles.
Merge pull request #6572 from RosettaCommons/roccomoretti/resize_ligand_puzzles
[Foldit] Some initial baseline reworking to help support multiple ligands/resizing.
The list of ligand designable positions in the SessionConfig is a mal-feature. If the pose changes size (e.g. if there's insertions or deletions, or if we fragment the designable ligands), then this numbering can become wrong. The proper place for this information is the ReferenceInfo, which gets properly resized when the structure is resized. We already have this info present, so make sure we use it.
This also cleans up a bunch of direct usage of the ReferenceInfo variables, in favor of the get_ligand_designable_positions() utility function.
(No release notes: internal cleanup only.)
Merge pull request #6563 from RosettaCommons/roccomoretti/load_with_secret_keys
[Foldit] Allow (scientist) loading of expired puzzles.
It would be helpful to be able for scientists to load expired puzzles. That's currently not possible through the standard client.
However, there is a secret key combination of CAS-N which was set to load specific Dojo puzzles. This functionality can be extended to download and launch any puzzle.
Attempting to launch an old-server puzzle (that is, a puzzle without file contents) will result in a crash -- as this is a secret key functionality, that should be okay.
(No release notes -- internal feature only.)
Merge pull request #6565 from RosettaCommons/roccomoretti/load_solutions_extend
[Foldit] Allow the LoadSaveDialog to be resized
The solutions list scrollable on the LoadSaveDialog was frustratingly small for me. Being able to extend it when Foldit is fullscreen will make it much easier to browse solutions (particularly the shared-with-scientist solutions).
This PR adds a resize handle to the dialog which allows you to extend the size of the dialog.
It also contains a bugfix for the Resizable widget. Occasionally (particularly when you're moving around the main window size) it looks like it could get stuck thinking that the panel size is below the minimum size, and refusing to resize -- even if the resizing would make the panel bigger. The new code in Resizable should stop that from happening.
RELEASE NOTES: Add the ability to resize the Open/Share Solutions panel.
Merge pull request #6566 from RosettaCommons/roccomoretti/fix_load_save_left_panel
[Foldit] Fix issues with the Delete and Load buttons for server solutions in LoadSaveDialog
The Load button when applied to server solutions invoked the downloader, but didn't wait until it was done before launching -- this adds a callback which encapsulates the post-download activity.
The issue with the delete button leaving results was a bit more involved. From my understanding, the issue is that the Boinc object cache for shared solutions is being updated by the server call, but removed entries are never cleared. This updates the server code to remove entries which are no longer on the server.
RELEASE NOTES: Fix Load & Delete buttons for Server Solutions in the Open/Share Solutions panel
Merge pull request #6574 from RosettaCommons/roccomoretti/linux_exit
[Foldit] Fix segfaults during Linux shutdown.
I've noticed segfaults when debugging on Linux, and have tracked it down to trying to access various bits of data during program destruction.
Address this by turning off the Server prior to destroying everything, and avoid re-calling the application shutdown code from within the application shutdown procedure.
RELEASE NOTES: Address Segfaults during closing the program on Linux
Merge pull request #6571 from RosettaCommons/roccomoretti/fix_copy_to_clipboard
[Foldit] Fix copy-to-clipboard error.
The reported copy-shared-solution-to-clipboard crash wasn't actually fixed by my previous fixes (PR #6559).
Instead, it looks like the issue is that the publish_solution API endpoint doesn't send back a JSON 'object', but instead sends an empty string. Be robust to this condition (where the API doesn't actually send anything back).
RELEASE NOTES: Fix issue with copy-to-clipboard in shared solutions panel.
Merge pull request #6557 from RosettaCommons/roccomoretti/add_smiles_to_ligand_prop
[Foldit] Add Formula and SMILES to ligand properties panel
There's been some persistent requests to get better information about the formula (number of atoms) and the SMILES string of the ligand. This PR re-works the Small Molecule Properties panel to add that information.
It also removes the need to click boxes to see the values. (With a little rewrite, actually computing properties does not add appreciably to time.)
Being able to export the SMILES for a ligand has also been a request. As a knowledgeable and determined person would be able to create a SMILES string from the visualization, there's no particular harm in allowing them to use a button to copy it to a clipboard. (The ability to input SMILES strings may be different.)
While I was at it, I also fixed an issue where the ligand properties and view panels were created overlapping.
RELEASE NOTES: Add ligand formula and SMILES string to Small Molecule Properties panel.
Merge pull request #6555 from RosettaCommons/roccomoretti/add_ligand_smiles_field
[Foldit] Add SMILES to the list of short data uploaded.
This will allow better data processing on the server-side. ( This is the client support for crowdgames/foldit-server-rails#674 )
This is tested to be robust to both proposed and existing Foldit server code. (With current server code, the extra payload is just ignored.)
RELEASE NOTES: (None)
Merge pull request #6559 from RosettaCommons/roccomoretti/fix_share_network_issues
[Foldit] Make server interactions more robust.
A traceback from [a bug report](https://fold.it/forum/bugs/copy-to-clipboard-crashes-foldit), indicates that there may be non-Foldit exceptions being thrown from Server calls. We should be able to catch that, and make sure that there's a handleable error which won't cause the client to crash.
(Note that due to indentation changes, the diff is more complicated that it really should be.)
RELEASE NOTES: Fix error handling in server interaction (copy-to-clipboard crashes)