Rxn Chemistry Tools

Configuration File: rxn_chemistry_tools.json Tool Type: Local Tools Count: 2

This page contains all tools defined in the rxn_chemistry_tools.json configuration file.

Available Tools

RXNChemistry_predict_reaction (Type: RXNChemistryTool)

Predict the most likely product of a chemical reaction using IBM RXN for Chemistry (ML forward re…

RXNChemistry_predict_reaction tool specification

Tool Information:

  • Name: RXNChemistry_predict_reaction

  • Type: RXNChemistryTool

  • Description: Predict the most likely product of a chemical reaction using IBM RXN for Chemistry (ML forward reaction prediction). Input is reactant/reagent SMILES joined by ‘.’ (e.g. ‘BrBr.c1ccc2cc3ccccc3cc2c1’); output is the predicted product SMILES with a confidence score. Submits an async job and polls until complete. Requires a free API key in the RXN4CHEMISTRY_API_KEY environment variable (register at https://rxn.app.accelerate.science).

Parameters:

  • operation (string) (optional) Operation type (fixed: predict_reaction).

  • reactants (string) (required) Reactant and reagent SMILES joined by ‘.’ (dot-separated). Example: ‘BrBr.c1ccc2cc3ccccc3cc2c1’.

  • ai_model (string) (optional) Optional IBM RXN AI model id for forward prediction. Defaults to ‘2020-08-10’.

  • project_id (string) (optional) Optional IBM RXN project id to run the prediction under. If omitted, the first available project is used (or one is created).

  • poll_interval (number) (optional) Optional seconds between status polls (default 5).

  • max_wait_time (number) (optional) Optional total polling budget in seconds (default 60).

Example Usage:

query = {
    "name": "RXNChemistry_predict_reaction",
    "arguments": {
        "reactants": "example_value"
    }
}
result = tu.run(query)

RXNChemistry_predict_retrosynthesis (Type: RXNChemistryTool)

Predict retrosynthetic routes for a target molecule using IBM RXN for Chemistry (ML retrosynthesi…

RXNChemistry_predict_retrosynthesis tool specification

Tool Information:

  • Name: RXNChemistry_predict_retrosynthesis

  • Type: RXNChemistryTool

  • Description: Predict retrosynthetic routes for a target molecule using IBM RXN for Chemistry (ML retrosynthesis). Input is the target product SMILES (e.g. aspirin ‘CC(=O)Oc1ccccc1C(=O)O’); output is a list of predicted precursor routes, each with a confidence/score. Submits an async job and polls until complete. Requires a free API key in the RXN4CHEMISTRY_API_KEY environment variable (register at https://rxn.app.accelerate.science).

Parameters:

  • operation (string) (optional) Operation type (fixed: predict_retrosynthesis).

  • product (string) (required) Target product SMILES to retrosynthesize. Example: ‘CC(=O)Oc1ccccc1C(=O)O’ (aspirin).

  • ai_model (string) (optional) Optional IBM RXN retrosynthesis model id. Defaults to ‘2020-08-10’.

  • max_steps (integer) (optional) Optional maximum number of retrosynthetic steps to explore.

  • project_id (string) (optional) Optional IBM RXN project id to run the prediction under. If omitted, the first available project is used (or one is created).

  • poll_interval (number) (optional) Optional seconds between status polls (default 5).

  • max_wait_time (number) (optional) Optional total polling budget in seconds (default 60).

Example Usage:

query = {
    "name": "RXNChemistry_predict_retrosynthesis",
    "arguments": {
        "product": "example_value"
    }
}
result = tu.run(query)