Spaya

Spaya is Iktos's AI-driven retrosynthesis platform designed to transform target compounds into commercially available starting materials in seconds. Built for medicinal chemists and synthetic chemists in drug discovery and research, Spaya enables users to explore all plausible synthetic routes in real-time, making the process of synthetic planning significantly faster and more efficient than traditional approaches.

At its core, Spaya employs a unique data-driven methodology that combines reinforcement learning and Monte Carlo Tree Search (MCTS) to identify new synthetic routes at unprecedented speed. This approach, paired with a template-based neural network trained on 6 million reactions, positions Spaya as one of the fastest retrosynthesis platforms available on the market.

Key Platform Capabilities

• Real-time retrosynthesis: Delivers retrosynthetic routes in seconds, enabling rapid exploration of synthetic strategies.
• Reaction template engine: A template-based neural network trained on 6 million reactions underpins route generation.
• Chemo-selectivity: Protection and deprotection strategies are incorporated to enable chemo-selective synthesis planning.
• Regio-selectivity: A dedicated neural network predicts the most probable regioisomer for each transformation.
• Chirality handling: Stereocenters are addressed using a chiral pool approach, drawing from chiral building blocks.
• Proprietary ranking: A proprietary scoring function ranks and identifies the top synthetic routes from all plausible options.
• MCTS algorithm: Monte Carlo Tree Search rapidly propagates retrosynthesis routes back to commercially available compounds.

Customizable Synthesis Constraints

• Access to partnerships with over 35 commercial building block providers.
• Support for over 2,000 named reactions.
• Ability to impose or forbid specific intermediates and substructures.
• Option to specify the first bond disconnection in a route.
• Filtering and customization by number of steps, provider pricing, and delivery times.

How Spaya Works

1. Draw a molecule directly in the interface or enter a SMILES string.
2. Select preferred options including providers, named reactions, intermediates, substructures, and disconnection preferences.
3. Browse results using a proprietary scoring system that identifies top routes, provides references for each route, and allows easy export of information.

Spaya API Suite

• High-throughput scoring: Input up to hundreds of thousands of molecules and receive instant synthetic feasibility scores.
• RScore: Iktos's proprietary score of synthetic accessibility for each molecule.
• Number of Steps: Returns the step count of the shortest identified route.
• RSpred Score: A fast estimation of synthetic feasibility that predicts RScore without running a full retrosynthesis.
• Seamless integration into existing workflows via platforms such as KNIME, Pipeline Pilot, Jupyter Notebook, and others.

Customization and Retraining

• Spaya can be retrained within a matter of weeks using proprietary reaction data, custom building blocks, intermediates, and internal inventory.
• This allows organizations to tailor the platform's predictions to their specific chemistry and compound libraries.

Spaya is available on AWS with autoscaling or can be deployed on a private VPC for enterprise environments. Iktos has also established a strategic partnership with Elsevier, integrating Spaya's AI technology with Reaxys, the world's largest chemistry database, to create advanced predictive retrosynthesis and synthetic accessibility tools. Training resources including tutorials, webinars, and publications are available to support onboarding and continued use.