Vision Arcanum logo
Ink2Mol onlineAPIBlog

InkToMolecule beta  User Guide

OneNote Interface

After installation of InkToMolecule add-in, a new button should appear on the Draw tab of the ribbon (see Figure 1).

Ink to Molecule button
Figure 1. "Ink To Molecule" button on the ribbon.


In order to recognize a hand-drawn chemical structure:
  1. Make sure that the Draw tab of the ribbon is active (see Figure 1).
  2. Select a structure that you want to recognize using Lasso Select tool.
  3. Make sure that the selection is treated as drawing: right click on the selection, then in the context menu pick "Treat Selected Ink As" 🠞 "Drawing" (see Figure 2 bellow).
  4. Press "Ink to Molecule" button on the ribbon.
  5. Ink to Molecule dialog should appear (see next section).

Treat as drawing
Figure 2. Treat selected ink as drawing.

Ink to Molecule Dialog

Note that InkToMolecule add-in uses our central recognition server and thus requires the Internet connection to work. The amount of time it takes to receive the server response highly depends on your network connection. A waiting indicator is displayed at the top of the dialog while waiting for the response.

Once the response is received, you should see the recognition preview in the central part of the dialog (see Figure 3). Also check the input drawing to make sure that it matches the intended selection.

Ink to Molecule dialogFigure 3. Ink to Molecule dialog.

You can switch between two modes of the recognition result. See Figure 4 for an example.

  • In the "Original structure" mode the result is preserved as close as possible to the input drawing.
  • In the "Clean structure" mode the coordinates of atoms are recomputed and abbreviations are expanded.

Recognition result modes
Figure 4. Recognition result modes.


In order to save the recognized structure as SDF file, press "Save SDF as..." button. Make sure that the desired recognition result mode is selected.

If you notice a mistake in recognition, please, press "Report a mistake" button to let us know.

In order to replace the recognized hand-drawn structure by a generated image, press "Insert" button. Make sure that the desired recognition result mode is selected. The dialog will close.


Which OneNote should I use?

Although Microsoft OneNote is available for various platforms, InkToMolecule add-in currently works only with OneNote 2016 for Windows. Curiously enough, on Windows 10 two completely different versions of OneNote can be used simultaneously (source):

  • OneNote (formerly called “OneNote 2016”), the free desktop app which runs on all supported versions of Microsoft Windows and which is part of Office 2019 and Microsoft 365.
  • OneNote for Windows 10, the OneNote app which is pre-installed on all editions of Windows 10, and which can also be downloaded for free in the Microsoft Store.
You should use the first version (called "OneNote"), since the second version (called "OneNote for Windows 10") does not support third party add-ins yet.

After installation the "Ink To Molecule" button does not appear.

First, make sure that you are checking the correct version of OneNote (see "Which OneNote should I use?" above).

If OneNote was running during the installation, try restarting it.

Input drawing shown in the dialog looks nothing like my selection.

OneNote automatically attempts to recognize any drawing as handwritten text. Since chemical structures typically combine graphic elements and labels, they tend to be treated as a mixture of drawings and texts. This makes it surprisingly difficult for an add-in to put the entire drawing together.

Make sure that whole your selection is treated as drawing: right click on the selection, then in the context menu pick
"Treat Selected Ink As" 🠞 "Drawing" (see Figure 2 above).

Recognition result is not correct.

We are the first to admit, that the current state of recognition accuracy leaves much to be desired. However, we can assure you that it does not reflect the final quality that we are striving to achieve. During the open beta we are going to collect usage data that will help us to improve our models and to provide better recognition accuracy.

We encourage you to give it a chance, and when you notice a recognition mistake, please, press the "Report a mistake" button to let us know.