Chem 1.11.0 (2024-07-24): docs, release notes
Dependency: datagarok-api >= 1.20.0
Bug Fixes
- GROK-16328: Chem: R-Group analysis: Broken styles
- GROK-16329: Chem: R-Group analysis: Error when none R groups were found
- GROK-16338: Chem: Chemical space: Reseting marker type to circles
- Chem: Scaffold Tree: Fix representation when the viewers are stacked
- Chem: Chemprop: Fix applying the model to different datasets
- Chem: Fix substructure filter style
- Chem: Do not render molecules with invisible width
2 Likes
Charts 1.3.3 (2024-08-08): docs, release notes
Bug Fixes
-
#2954: Sunburst: Unsupported columns are not filtered out in column selector
1 Like
Hi,
Is it possible to chemprop from datagrok for compound properties prediction?
If yes, are there any specific settings to use chemprop from datagrok?
Thanks,
Taka
Chem 1.11.2 (2024-08-12): docs, release notes
Features
- Chem: Chemprop: Remove redundant calls in the core and move it fully to package
- Chem: MMPA: Add meaningfull dataset for demo
- Chem: MMPA: Restrictions and CPU fallback
- Chem: MMPA: Compensate filtering by cutoff
- Chem: MMPA: Add reflection for selection
- Chem: MMPA: Add pairs filter, expose filters object so it can be used in any tab, minor improvements
- Chem: Update demos in order to be shown in the Browse
Bug Fixes
- Chem: MMPA: Fixes for cliffs tab and lines improvements
-
#2964: Scaffold Tree: āSelection to filterā reset immediately for some data if scaffold tree is present
1 Like
Hi Taka,
Yes, Chemprop is fully available from the Chem package. To use it for compound properties prediction, follow these steps:
- Open the molecular dataset of your interest.
- Navigate to ML > Models > Train Modelā¦
- Select table, column with the values you want to predict, and the feature column (in this case, it will be the column containing smiles or molblocks).
- You will then be presented with a list of suitable engines. Among them, you will find the Chemprop model engine.
- Simply choose Chemprop, configure the model settings, and run the training.
Once your model is trained, you can apply it by going to ML > Models > Apply Modelā¦.
If you have any questions, donāt hesitate to contact me
Best regards,
Oleksandra
1 Like
EDA 1.1.34 (2024-08-19)
This release improves Datagrok predictive modeling tools:
- Updated Multivariate Analysis
- Improved efficiency of the PLS regression
- Fixed the memory issue in the linear regression
Also, XGBoost is available:
- Open dataframe
- Go to Top Menu > ML > Models > Train Modelā¦
- Set
Predict
and Features
- Select
Eda: XGBoost
in the Model Engine
field
- Tune settings to get better efficiency
- Click on the
Save
button to save the trained model
- Apply the trained model via Top Menu > ML > Models > Apply Modelā¦
Check interactivity of Datagrok predictive tools
1 Like
Charts 1.3.4 (2024-08-23): docs, release notes
Bug Fixes
-
#2994: Charts | Sunburst: Double click on the viewer resets the view
-
#2992: Charts: Sunburst does not select/filter on empty category
- Charts: Sunburst: Update emphasis in order not to fade out other data
- Charts: Sunburst: Use only categorical columns for building the hierarchy
-
#2979: Charts: Sunburst is showing default columns instead of selected ones after applying layaout in some cases
- Charts: Demo: Fix globe viewer doesnāt show points
1 Like
Chem 1.11.4 (2024-08-23): docs, release notes
Features
- Chem: MMPA: Filter fragments by significance
- Chem: MMPA: Project and layout saving
- Chem: MMPA: Demo loaded not calculated
- Chem: Removed r-groups settings download from init
Bug Fixes
- Chem: MMPA: Fixed property panel on cliff line click
-
#2983: Chem: Fix tooltip styles for structure columns
- GROK-16493: MMPA: Fixed Summary viewer doesnāt show the values
1 Like
Chem 1.11.5 (2024-08-27): docs, release notes
Bug fixes
- Add tableView to runChemSpace for reduce dimensionality ui options
-
#3004: Chem: Scaffold tree is based on wrong table by default in some cases
- GROK-15938: Chem | Scaffold Tree: Array length differ
- Chem: MMPA: Fix demo
1 Like
Diff Studio 1.1.1 (2024-09-11): docs
- Browsing: Templates/Examples/Recent models integration to the browse tree
- Added tests & benchmarks: solvers performance & correctness, models features (basic structure, loops, updates, etc.)
- Templates/Examples: updated the basic template, added the Pollution model
- Float64 columns usage implementation
- Improved the main app start
1 Like
Diff Studio 1.1.2 (2024-09-12): docs
This release implements two models:
Explore them at Apps > Model Catalog. Also, check demo applications at Apps > Demo > Compute.
1 Like
EDA 1.2.1 (2024-09-12): docs
Added tests & benchmarks for missing values imputation using k-NN
1 Like
Diff Studio 1.1.3 (2024-09-25): docs
This release provides the use of lookup tables. Check it:
- Go to Apps > Diff Studio > Examples
- Run Bioreactor
1 Like
Diff Studio 1.1.4 (2024-09-30): docs
This release enables the use of lookup tables in Sensitivity Analysis and Parameters Optimization.
1 Like