AI-Powered Molecular Property Prediction
Screen promising molecules before laboratory testing
Developing a new dye, fluorescent probe, organic emitter, UV absorber, or light-harvesting molecule can require many rounds of design, synthesis, and testing.
Our molecular property prediction service helps research teams evaluate candidate molecules earlier. We use deep learning to estimate important optical and photophysical properties from molecular structure and solvent information. Agentic AI supports the wider workflow by organising inputs, applying screening rules, comparing candidates, and preparing clear reports.
This allows researchers to review a larger number of molecules and create a more focused shortlist for synthesis, spectroscopy, or advanced computational studies.
What We Predict
The service can estimate seven important properties:
- › Absorption maximum
- › Emission maximum
- › Absorption bandwidth
- › Emission bandwidth
- › Extinction coefficient
- › Photoluminescence quantum yield
- › Emission lifetime
These properties are useful when studying molecular colour, fluorescence, brightness, spectral behaviour, and excited-state dynamics.
Why the Solvent Matters
The optical behaviour of a molecule can change depending on its surrounding environment. The same chromophore may show different absorption or emission properties in different solvents.
Our model therefore considers both:
- The molecular structure of the chromophore
- The molecular structure of the solvent or supported environment
This makes the screening process more relevant to real experimental conditions than a structure-only prediction.
Some laboratory factors, such as temperature, concentration, pH, aggregation, oxygen level, and material morphology, may still affect the final result and should be considered during experimental validation.
How It Works
Share your molecular candidates
Provide the SMILES structures, intended solvents, and any target property ranges.
We review the inputs
The submitted structures are checked for invalid or unsupported molecular information.
Deep learning predicts the properties
The model represents molecules as graphs of atoms and bonds. It analyses the chromophore and solvent together and estimates the seven properties.
Agentic AI supports screening
Agentic AI helps organise the results, apply your selection criteria, compare molecules, and rank candidates based on your research goals.
You receive a clear report
The final output can include predicted properties, candidate rankings, input warnings, comparisons with available experimental values, and a recommended shortlist.
Research Areas
This service can support projects involving:
- ✓ OLEDs and organic emitters
- ✓ Fluorescent and bioimaging probes
- ✓ Solar dyes and light-harvesting materials
- ✓ UV absorbers and photoprotective molecules
- ✓ Chemical sensors
- ✓ Photoinitiators
- ✓ Photocatalyst research
- ✓ General chromophore and fluorophore development
What You Receive
Depending on the project, we can provide:
- Predictions for all supported properties
- A ranked list of molecular candidates
- Screening based on your target ranges
- Comparison with available experimental data
- Notes on unusual or uncertain predictions
- A technical report or structured spreadsheet
- A recommended shortlist for further research
Start With a Screening Pilot
A small pilot is a practical way to evaluate the service for your chemical space.
You can submit:
- › 20–100 molecular structures
- › The intended solvent or environment
- › Available experimental values
- › Your preferred property ranges or screening goals
We will return:
The predictions, input checks, candidate ranking, and a clear summary of the most promising molecules.
Ready to Screen Your Molecular Candidates?
Use deep learning and agentic AI to narrow your search before committing to synthesis and laboratory testing.
Discuss a Screening Pilot