Synthesis Planning
Plan retrosynthetic routes for target molecules and assess building block availability using AI-powered analysis.
Synthesis Planning
Synthesis planning translates a target molecule into a practical laboratory procedure. This workflow uses AI-driven retrosynthetic analysis to decompose complex molecules into commercially available building blocks, then verifies reagent availability and estimates costs across global suppliers.
Overview
Retrosynthesis is the process of working backward from a target molecule to identify sequences of known reactions that lead to purchasable starting materials. AiZynthFinder uses Monte Carlo tree search guided by neural network expansion policies to explore millions of potential routes and rank them by feasibility, step count, and building block availability. Once routes are identified, a ZINC lookup tells you which building blocks are purchasable and from which vendor catalogs, so you can make informed make-versus-buy decisions.
Tool Chain
- AiZynthFinder -- AI-powered retrosynthetic route planning
- ZINC -- Purchasability and vendor catalog codes for building blocks
Prerequisites
- A valid SMILES string for your target molecule
- Familiarity with basic organic chemistry reaction classes
- A platform account with access to synthesis tools
Step-by-Step Guide
Define the target molecule
Enter the SMILES representation of your target compound into AiZynthFinder. You can paste SMILES directly, draw a structure, or select a compound from a previous job such as a de novo design or virtual screening hit. Verify the structure preview matches your intended molecule before proceeding.
Configure retrosynthesis parameters
Select the expansion policy (default works well for drug-like molecules) and set the stock repository. The stock repository determines which building blocks are considered purchasable. The bundled configuration defines a zinc stock; see AiZynthFinder API for the parameters a job accepts. Set the maximum number of search iterations (higher values explore more routes but take longer) and the maximum route depth (typically 6-10 steps).
Run retrosynthetic analysis
Submit the job and monitor progress in real time. AiZynthFinder typically completes in 2-15 minutes depending on molecular complexity and search depth. The algorithm explores a tree of possible disconnections and prunes branches that lead to unavailable or impractical intermediates.
Review and rank routes
Examine the returned routes in the interactive tree viewer. Each route shows the sequence of reactions from target to building blocks, with confidence scores for each step. Key metrics to evaluate include total step count, estimated overall yield, number of commercial building blocks available, and the presence of well-precedented reaction types.
Verify building block availability
For your preferred route, ask Ivy in chat for a ZINC lookup of each building block. It returns ZINC IDs and vendor catalog codes; Ivy can then run a vendor price search for indicative prices, which may be out of date, so confirm quotes with the vendor. Flag any building blocks with long lead times or high costs that might warrant route revision.
Export and document
Export the selected route as a PDF or JSON artifact. Include route diagrams, building block lists with catalog numbers, and estimated costs in your project records for handoff to medicinal chemistry teams.
Expected Outputs
- Ranked retrosynthetic route trees with confidence scores per step
- Building block lists with CAS numbers and SMILES
- Purchasability and vendor catalog codes from ZINC, plus indicative prices where a vendor publishes them
- Exportable route diagrams in PDF and JSON formats
- Cost estimates for reagents and starting materials
Tips and Best Practices
- Start with default parameters for an initial survey, then tighten constraints (fewer steps, specific vendors) in subsequent runs.
- Use the combined stock repository for the first analysis to maximize the chance of finding viable routes, then narrow to preferred suppliers.
- For complex natural product-like molecules, increase the search iteration limit to allow deeper exploration.
- Compare at least three alternative routes before committing to synthesis. The top-ranked route by algorithm score may not be the most practical given your lab capabilities.
- Cross-reference building block prices with bulk discount tiers if you plan to synthesize at scale.
- Document all route decisions with rationale to support reproducibility and regulatory submissions.
Run it from a prompt
The same goal as a copy-and-adapt recipe you can paste into a chat:
- Hit to lead — plans a route for a docked, ADMET-screened hit and checks vendor availability in one prompt.
Was this page helpful?