Find diverse molecules with
similar activity. Fast.
Search Quality Benchmarks
BIOPTIC outperforms existing methods when tested on DUD-E and LIT-PCBA virtual screening benchmarks.
* LIT-PCBA virtual screening benchmark
** NEF@1% - Normalized Enrichment Factor at 1%
*** Virtual Screening with Gnina 1.0 ↗
Activity, Not Structure
BIOPTIC's Deep Learning fingerprinting function is trained on large drug-target bioactivity libraries. Simply input a known ligand or a promising candidate and BIOPTIC will return a diverse set of active, target-compatible variants.
Case study
Finding GLP-1 inhibitors with diverse structure
Using BIOPTIC partner ran Pfizer PF-06882961 (Danuglipron) candidate against a library of 5.6B compounds. Top 40 hit was a known compound in early clinical development program of another pharmaceutical company with activity value of 0.889. As you can see from the structure it is vastly different from Pfizer candidate.
Pfizer PF-06882961 Candidate (Danuglipron).
Top 40 Hit Molecule (Research compound by another company).
Key Features
Activity-Based Search
Focus on predicted molecular activities to identify potential hits with higher precision.
Unlimited Database Size
Seamlessly search through databases of any size without compromising on speed or accuracy.
Scaffold Hopping
Retrieve a diverse set of molecules for hit identification and drug novelty.
Experience the Future
Revolutionize your drug discovery journey with our AI Engine. Discover, collect, and explore diverse molecules with unprecedented speed and efficiency. The future of drug discovery is at your fingertips.
Revolutionizing research
Discover, compare, analyze. Our upcoming AI features will greatly enhance molecular analysis and safety capabilities.
Protein-Molecule Interactions
- Molecule to Protein Mapping: Innovative tool to find molecules linked or interacting with a chosen protein.
- Protein to Molecule Mapping: Identify proteins that potentially interact with a given molecule, enhancing the understanding of drug-target relationships.
Off-Target Activity
- Automated Filtering: Assess the safety profile using predicted activities to our carefully curated list of proteins causing the majority of adverse effects.
- Custom Molecule Analysis: Opportunity to upload and analyze your molecule for a detailed safety assessment against our protein database.
Dual Molecule Matching
- Simultaneous Comparison: Unique capability to concurrently compare two molecules, offering insights into their collective interactions and synergies.
Custom Protein Profiling
- Personalized Analysis: Flexibility to create safety profiles using a specific list of proteins, tailored for unique biomarkers or therapeutic targets.
FAQs
BIOPTIC is a ligand-based molecular search engine utilizing a deep neural network to represent molecules through activity-expressing fingerprints. It's designed to find compounds with similar activity properties to those in your hit list, expanding your molecular discovery possibilities.
BIOPTIC uses a deep learning model pre-trained on the ZINC molecular database and fine-tuned with data from bioactivity databases like ChEMBL, BindingDB, and PDBbind. It accepts a molecule's SMILES representation and outputs a 64-dimensional activity fingerprint, enabling it to find molecules with similar biological activities.
BIOPTIC stands out for its activity-based similarity search, focusing on activity-centric features rather than structural similarities. This approach allows for discovering compounds across different scaffolds, making it a powerful tool for novel drug discovery.
Yes, BIOPTIC is designed to handle ultra-large chemical libraries. Its scalable GPU-based cloud infrastructure enables the processing of extensive datasets efficiently, making it suitable for high-throughput molecular searches.
BIOPTIC's performance is benchmarked against well-known datasets like DUD-E and LIT-PCBA. These benchmarks assess the tool's ability to prioritize active compounds, ensuring high-quality, relevant search results.
Currently, BIOPTIC supports the SMILES format for molecule input. This format is widely used in chemical informatics for representing molecular structures in a readable text form.
Absolutely, BIOPTIC is ideally suited for hit expansion. It helps in identifying novel compounds with similar activities to known hits, thereby aiding in the drug discovery process.
BIOPTIC calculates the cosine distance between the activity fingerprints of molecules. This similarity score, ranging from 0 to 1, helps in assessing the closeness of molecular activity profiles.
BIOPTIC is designed to work with vast chemical libraries, and its scalable infrastructure can handle extensive datasets without a significant compromise in speed or accuracy.
BIOPTIC differs from traditional molecular docking tools as it focuses on activity-based similarities rather than structural similarities. This unique approach allows for the discovery of compounds that might be overlooked by conventional docking methods, offering new avenues in drug discovery.
Still have questions?
Feel free to text us
Partner with Bioptic
Expanding the scope of your research.