Reaction Templates

For large factorial datasets, reaction templates are a mechanism to quickly enumerate a set of reactions based on a spreadsheet of values.

Step 1: Create a template reaction

Use the ORD Reaction Editor to create a new dataset and fill out a single reaction (feel free to enter dummy values for fields that vary in the dataset). The following videos from the ORD YouTube channel are useful guides to creating datasets and reactions using the editor:

Here’s an example reaction.

Step 2: Mark the variable fields in the template

Use the ORD Reaction Editor to turn the single reaction into a template reaction. This YouTube tutorial on dataset enumeration shows how to do this. Then use the graphical interface to select and label any fields that vary in the dataset.

Open a text editor such as Notepad (Windows) or TextEdit (Mac) and replace the template fields with variable placeholders. Each variable name should match a column in the data spreadsheet and start and end with $. For instance:

...
products {
  identifiers {
    type: SMILES
      value: "C"  <-- Change "C" to "$product_smiles$" (with quotes).
  }
  is_desired_product: true
  measurements {
    analysis_key: "19f NMR of crude"
    type: YIELD
    uses_internal_standard: true
    percentage {
      value: 50.0  <-- Change 50.0 to $product_yield$ (without quotes).
      precision: 4.8
    }
  }
...

Here’s the templated version of the earlier example.

Step 3: Prepare the accompanying spreadsheet

The spreadsheet can be a CSV or Excel file. There should be a column for each of the variables defined in the previous step (the $ markers are not required). Here’s an example spreadsheet.

Important

Please note that the reaction enumerator in the new online ORD Reaction Editor needs the source data to be provided as a CSV file with semi-colon separated values (not comma separated).

Some datasets may not include all reaction components in every example. In these cases, the corresponding cell(s) in the spreadsheet should be left empty (see templating.py for details):

If after templating, any identifier does not have a defined value, remove the identifier.
If a compound doesn’t have any identifiers, remove that compound.
If a compound amount is undefined/NaN, remove that compound.
If an input has no components, remove that input.

Step 4: Enumerate the dataset

Use the ORD Reaction Editor and select the “Enumerate” button on the main page. Follow the instructions to select your template reaction, upload your source data as a semi-colon separated CSV file, and check the matching between columns and template fields. Click the “Create” button to build the dataset. This YouTube tutorial on dataset enumeration shows how to do this.