Idea
A protein surface retrieval platform leveraging shape and electrostatic potential for accurate molecular similarity search in drug discovery and bioinformatics.
Research Paper
Core Innovation
This paper introduces a benchmark and evaluation of protein surface retrieval methods that integrate electrostatic potential with molecular shape, demonstrating superior retrieval performance. It uniquely shows that combining these descriptors improves accuracy even for protein classes with limited data, advancing beyond shape-only approaches.
Market Size (TAM)
$2–10B TAM for computational biology and drug discovery platforms; $1–2B SAM from pharmaceutical and biotech companies adopting advanced protein analysis tools. Driven by increasing demand for precision medicine and AI-powered molecular modeling.
Potential Customers & Pain Points
- Pharmaceutical companies needing precise protein similarity search
- Bioinformatics researchers requiring robust molecular surface descriptors
- Structural biologists analyzing protein functions
- AI developers lacking comprehensive protein retrieval benchmarks
Business Model
Subscription-based SaaS platform offering API access and custom analytics for protein surface retrieval; enterprise licensing for pharmaceutical and research institutions.
Competitive Landscape
- Schrödinger
- OpenEye Scientific
- Cresset
Implementation Challenges
- Integration complexity of electrostatic data with shape models
- High computational cost for large-scale protein datasets
- Adoption resistance due to existing shape-only tools
Validation Strategy
- Benchmark platform with public protein datasets and metrics
- Pilot collaborations with pharma partners for real-world testing
- Iterative improvement based on user feedback and retrieval performance
Research Paper Overview
SHREC 2025: Protein surface shape retrieval including electrostatic potential
Summary
This SHREC 2025 track evaluated 15 methods from 9 teams on retrieving protein surfaces using shape and electrostatic potential descriptors across 11,555 proteins. Methods combining electrostatic potential with molecular surface shape outperformed others, especially for classes with limited data, highlighting the value of integrating multiple molecular surface descriptors for improved retrieval accuracy.