Projects and files
What a project file holds, how to save and reopen one, and where the program keeps its other files.
A project gathers the datasets and workflows of a session in one .gqproj file. It is the way to stop work in the evening and continue the next day, or to give a colleague the whole analysis instead of a collection of files.
Saving and opening a project
- File > Save project (Ctrl+Shift+S, or Save in the Project group of the Home tab) writes the session into a project file; the first time it asks for a name. File > Save project as writes a copy under a new name. Both are available as soon as a dataset or a document is open.
- File > Open (Ctrl+O) opens a project. A project can also be dragged onto the window, chosen from the recent files, given on the command line, or opened by a double click in Explorer once the file association is switched on in Settings. If GeinQSAR Studio is already running, the file is passed to the open window.
- The save is written in one step, so a project file is never left half-written by a crash or a full disk.
What a project holds
| Stored in the project | Not stored |
|---|---|
| Every dataset that has rows, with all its columns, hidden ones included: as a table, or as structures with coordinates when the dataset holds 3D structures. | Results of workflow runs, trained models and dashboards: they are recomputed. |
| Every workflow, with its parameters, canvas groups and notes. | Figures and reports: they are separate files. |
| A list of which workflows had results when the project was saved, so that they can be offered. | The window layout, the theme and other settings: they belong to the program, not to the project. |
| The name, author, creation and modification dates, the program version and a checksum for every entry. | External files that a workflow reads by path, unless they were brought in as datasets. |
Opening a project computes nothing
Opening a project restores its datasets and opens its workflows on the canvas, fitted to the view, and that is all. The dashboards are empty until you ask for them. A project opens at once, however long its models took to fit, and no output file is rewritten behind your back.
The workflows that had results when the project was saved are offered rather than run. A notification that stays until it is answered says One workflow had results, or names the number when there are several, and explains that a project file stores the workflows and not their results, so running them again recomputes the models, the dashboards and the output files. Run it, or Run them, starts them one after another; Workflow > Recompute the project's results does the same at any time while they are still pending. Because every random step of a workflow is seeded, a rerun gives the same numbers as before.
A model that is wanted without recomputation should be saved as a file of its own: Save model on the 3D-QSAR dashboard, the model file of the 2D-QSAR wizard, or a Write model node. File > Open on a saved 3D-QSAR model shows its dashboard again without any workflow.
Integrity and compatibility
Each part of a project carries its own checksum. A file damaged in transfer or storage is refused when it is opened rather than read wrongly, and gqsar-cli project P.gqproj --verify checks every entry and lists the ones that do not match (see The command line). A project saved by a later version with a new major format version is refused with a message that names the version; a project of a later minor version opens, and parts this version does not know are skipped.
Unsaved changes
The program does not save automatically, and it does not reopen the documents of the previous session by itself: what is not saved in a project or a file is gone when the program is closed. With Ask before closing with unsaved changes on (the default), closing the program while a dataset has unsaved changes asks This project has unsaved changes. Close it anyway?
Recent files
The Recent list of the welcome page and File > Recent files show the 20 projects and datasets used last, with the number of compounds of each dataset. The lock icon on the welcome page pins an entry, so that it stays at the top and is never pushed off the list; the cross removes it. A file that has been moved or deleted stays in the list, greyed out and marked (missing). Clear the list in the menu removes every entry that is not pinned. The list is kept with the settings.
Where files go
| Location | What is there |
|---|---|
| The folder you choose | Projects, workflows, exported tables, models, reports and figures. The file dialogs start in the folder used last for each purpose, or in the fixed folders of Settings > Folders. |
Documents\GeinQSAR | The Dataset Builder's first suggestion for a new dataset; afterwards it offers the folder it saved to last. |
%APPDATA%\Geinforce\GeinQSAR | The application data folder (Tools > Open the data folder). |
settings.json | The settings and the recent-files list. |
layout-v2.json, window.txt | The arrangement of the panels and the window position. |
keybindings.json | Changed key bindings, if any. |
logs\geinqsar.log | The session log (Tools > Open the log folder). |
http-cache | The answers of ChEMBL and PubChem, so that a repeated download is fast and works offline. Searches and counts are asked again after 30 days. |
snapshots | Copies of datasets that the wizards and opened projects work from. |
Every file a dialog offers to write starts from a full path under Documents\GeinQSAR\Output, and a relative name you type yourself is resolved against that folder rather than against the folder the program was started from. A saved workflow keeps writing beside its own file. Settings describes the GQ_OUTPUT_DIR override.
Deleting the application data folder while the program is closed returns it to its factory state; projects and other files you saved elsewhere are not affected.
Sample data
The data folder of the installation holds the datasets used by the welcome page, the wizards and the templates, and the reference material of the validation:
| File | Content | Used by |
|---|---|---|
data/qsar3d/steroids.csv | The 31 corticosteroid-binding globulin steroids, the classical 3D-QSAR steroid benchmark of Cramer, Patterson and Bunce (1988), with the literature division into 21 training and 10 test compounds. | Welcome page CBG steroids; the 3D-QSAR templates and wizard |
data/chem/drugs.smi | About 200 well-known drugs and natural products, with formula, molecular weight and PubChem CID. | Welcome page Reference drugs |
data/geom/chiral_drugs.smi | 117 chiral drugs. | Welcome page Chiral drug pairs |
data/flow/ache_pic50.csv | 429 acetylcholinesterase inhibitors with pIC50 from ChEMBL (target CHEMBL220), pooled from 43 publications. | The 2D-QSAR, automatic model search, virtual-screening and library design templates |
data/flow/herg_class.csv | 554 compounds labelled hERG blocker (186) or non-blocker (368). | The classification template |
data/flow/DATASETS.md | How the two sets above were curated. | |
data/desc/chirality_cliffs.csv | Stereoisomer pairs with measured activities (34 compounds). | The chirality cliff template |
data/flow/templates | The eight workflow templates as .gqflow files. | |
data/qsar3d/New_Steroids_2D.csv | The 31 steroids with 2D descriptor columns. |
The welcome page buttons open the import dialog on the sample, so it can be checked before it is loaded. If a sample is missing, a message Sample not installed names the file that was looked for.