Method wordCloud doesn't show all the word clouds as it should, missing even the most frequent ones.

Example of such behavior can be found in word-cloud JS API example when choosing column site to show word clouds for.

Implement a word cloud viewer.

WordCloud library: https://github.com/ecomfe/echarts-wordcloud
Developing custom viewers: https://datagrok.ai/help/develop/how-to/develop-custom-viewer

Note that '203' on the Y axis is not shown since it's too close to '200'

The ability to view specified events (including time intervals) for the specified cohort on one chart with the X axis set to event date, aligned to the beginning of the study

Mark the quartiles Q1 and Q3 on the Histogram axe, highlight the area outside this range with filled rectangles. Set default filter markers to these values.

Integrate DrugBank with Datagrok, as it is currently realized for PubChem and Chembl.

https://go.drugbank.com
https://www.youtube.com/watch?v=k1NVdTRpYOM&t=181s (Datagrok: info panels for molecules)

1. Similarity & substructure search for compounds (as info panels)
2. Functions for similarity and substructure searches

Either DrugBank API, or complete downloaded database could be used.

Integration with ChEMBL

https://www.ebi.ac.uk/chembl/
https://chembl.gitbook.io/chembl-interface-documentation/web-services/chembl-data-web-services
https://www.ebi.ac.uk/chembl/api/data/docs

1. Info panels for molecules (substructure and similarity search)
2. Datagrok functions for the most popular web services
3. ChEMBL browser

Scripts supposed to be executed on the server have a special place in a package -- the /scripts directory. If script files are somewhere else, they may not be loaded properly into the platform. To prevent such cases, the location of files can be checked during the package publication step (grok publish).

• Ability to define complex criteria via the UI (ex: "subjects with at least two measurements of cholesterol > 200mg/dL")
• Reusable between analyses
• Shareable

Provides support for dealing with clinical data represented in the SDTM format

Sources: https://github.com/datagrok-ai/public/tree/master/packages/ClinicalCase

Implement a way to create view of DataFrame.
View creation function should accept BitSet / list of row indices along with list of column id's.

df = DG.FromJson({'a': [1, 2, 3, 4], 'b': [5,6,7,8]})
indices = [0,2,3] 
cols = ['b']
v = df.view(indices ,cols )

v => 'b'
         5
         7
         8

chem_error throws error: malformed molString for arguments that are seemingly good

`
Mrv1902 02241904102D

11 10 0 0 0 0 999 V2000
0.9818 0.5834 0.0000 Cl 0 0 0 0 0 0 0 0 0 0 0 0
0.9818 -0.5834 0.0000 Cl 0 0 0 0 0 0 0 0 0 0 0 0
0.3984 0.0000 0.0000 Pt 0 0 0 0 0 0 0 0 0 0 0 0
0.0286 1.3803 0.0000 H 0 0 0 0 0 0 0 0 0 0 0 0
-0.9818 -0.3699 0.0000 H 0 0 0 0 0 0 0 0 0 0 0 0
-0.9818 0.3698 0.0000 H 0 0 0 0 0 0 0 0 0 0 0 0
0.0285 -1.3803 0.0000 H 0 0 0 0 0 0 0 0 0 0 0 0
-0.1849 -0.5834 0.0000 N 0 0 0 0 0 0 0 0 0 0 0 0
-0.3984 -1.3803 0.0000 H 0 0 0 0 0 0 0 0 0 0 0 0
-0.1849 0.5834 0.0000 N 0 0 0 0 0 0 0 0 0 0 0 0
-0.9818 0.7969 0.0000 H 0 0 0 0 0 0 0 0 0 0 0 0
3 1 1 0 0 0 0
3 2 1 0 0 0 0
3 8 1 0 0 0 0
3 10 1 0 0 0 0
10 6 1 0 0 0 0
10 4 1 0 0 0 0
8 5 1 0 0 0 0
8 7 1 0 0 0 0
8 9 1 0 0 0 0
10 11 1 0 0 0 0
M END
`

In the google-map example map doesn't show.

If a dialog is set to be modal and full screen with .showModal(true), clicking outside of the dialog will result in closing it.

Example of such behavior can be found in JS API Examples: ui/dialogs/full-screen.js.

When trying to change string/numeric input value in ML -> Train model... after opening a feature selection menu, focus is lost and the value is not changed. It happens only once exactly after using the feature selection menu, trying to change the value again works as expected.

A collection of interactive tutorials

• Interactive hints
• Tracks (collections of tutorials)
  • Exploratory data analysis
  • Machine learning
  • Cheminformatics
• Badges

• subject enrollment chart
• basic statistics (numbers of subjects, distribution for SEX / RACE / AGE)

The package sources: https://github.com/datagrok-ai/public/tree/master/packages/Charts

This package will provide a TensorFlow.js-based predictive model engine that would be integrated with Datagrok's predictive modeling.

Here is an overview of the existing functionality:
https://youtu.be/JaJgxtHAb98?t=1417

When question mark is clicked in a dialog window, it doesn't lead anywhere even if helpUrl is set and valid.

Example of such behavior can be found in JS API Examples ui/dialogs/dialogs.js.

https://lifescience.opensource.epam.com/mri/

The package will provide file preview capabilities for DICOM, NIfTI, KTX, and HDR file formats.

Functionally, the implementation would be quite similar to the NGL Viewer package.

This will be used for the client-based multiple sequence alignments across our bioinformatics solutions when it's computationally feasible (examples: peptides).

@StLeonidas , which algorithms are we implementing exactly (ClustalV / ClustalW / ClustalOmega / etc)?

• Integrate OpenChemLib sketcher as one of the molecular sketchers
• Other cheminformatics functionality

The following line
grok.shell.add(grok.data.demo.wells());
makes Datagrok look like on screenshot (only on public.datagrok.ai)

Viewers like histogram and scatterplot use collaborative filtering. It will be useful to add the same feature to bar chart.

Examples:

Domains: CV -> Cardiovascular System Findings
Attributes: AEMODIFY -> Modified Reported Term

Ability to search for anything from the start screen, with the special support for the following:

• Widgets
• Functions
• Applications
• User-defined external apps (via iframe)
• Entities (connections, queries, etc)

• Move all summary row renderers out of the core into the "Sparklines" package
• Introduce the ability to customize row renderers

https://datagrok.ai/help/visualize/viewers/grid#row-summary-columns

It is useful to be able to render reference pre-calculated distributions on top of the real, calculated ones. It would make sense to keep them with the dataframe in column tags. This would enable many scenarios, such as constructing a dataframe on a server or in a Python script.

Once a reference distribution is defined for a column, then the histogram (and filter) would automatically pick it up and show.

I propose the following approach:

demog.weight.tags['.ref-distributions'] = `{
  'healthy': [2, 4, 45, 76, 45, 5, 5, 6, 2, 1],
  'sick': [3, 4, 56, 76, 45, 5, 5, 6, 2, 1],
}`;

Integration with PubChem

https://pubchem.ncbi.nlm.nih.gov/
https://pubchemdocs.ncbi.nlm.nih.gov/pug-rest

Info panels for molecules (substructure and similarity search)
Datagrok functions for the most popular web services
PubChem browser

Currently it is not possible to set int and float viewer properties to be in specific range and they all default to range from 0 to 100 on a slider.

It would be useful to set min, max and step options for int and float properties like this:
this.myIntProperty = this.int('myIntProperty', 0, {min: -10, max: 10, step: 2});
this.myFloatProperty = this.float('myFloatProperty', 0, {min: -1, max: 1, step: 0.2});

As users go through the tutorials, we'd like to track their progress and give rewards for their achievements. For example, they can be shown in a separate pane of the user's profile.

Integrate Ketcher as one of the molecule sketchers supported by the platform.

https://lifescience.opensource.epam.com/ketcher/index.html

Example of a similar integration with the OpenChemLib's sketcher: https://github.com/datagrok-ai/public/blob/master/packages/OpenChemLib/src/ocl-sketcher.ts

Existing prototype (works but does not synchronize with smiles): https://github.com/datagrok-ai/public/tree/master/packages/Ketcher

To open a dialog with the sketcher:
grok.chem.showSketcherDialog();

Running dynamic-loading example throws the following error in browser console: Uncaught (in promise) TypeError: Cannot read property 'name' of undefined.

In public it also breaks Datagrok style as shown in #72 and in dev it doesn't list handlers in view in contrast to public.

File preview

• KMZ
• KLM

Integration with OpenStreetMap

Wiki: Structure-Activity Relationship (SAR) is an approach designed to find relationships between chemical structure (or structural-related properties) and biological activity (or target property) of studied compounds.

We need to build an interactive analysis tool to help identify point mutations that cause large change in activity.

Input is a dataframe with the following columns:

1. peptide: aligned sequences (max length N, number of unique amino acids A)
2. activity: measured activity

Possible outputs:

An N*A dataframe where (n, a) element contains a measure of the change of activity caused by mutating the amino acid at position n to a.

Visualize a collection of peptides in 2-dimensional space, using t-SNE with the edit distance function. Implement Peptide space feature similar to Chemical space feature in Chem package:

• Implement an algorithm to calculate similarities for a pair of sequences (semtype - alignedSequence)
• Use SPE algorithm to get 2d map of peptide sequences
• Implement a new feature in the property panel of each column with alignedSequence values
• Implement embedding as a web worker to not slow down the main thread.
• Benchmark UMAP performance on dataset of 40k sequences.

Currently, we open SDF files via the server-based RDKit library, which is suboptimal. We can do it completely on the client side by either using OpenChemLib (see SDFFileParser - the link is not opening at the moment though), or parsing it manually.

https://github.com/datagrok-ai/public/blob/master/help/develop/how-to/file-handlers.md

The exported function should reside in the Chem package.

This package would read SQLite files as Datagrok's data frames. It would utilize WASM-compliled SQLite so that everything would happen inside the browser.

https://github.com/sql-js/sql.js/

A start page that contains widgets (annotated with the dashboard tag) that are dynamically discovered from the packages available to the current user.