Complex Structure Prediction (Protein-Protein/Peptide Interaction Modeling)
Multimeric AI Assembly 
Interaction Modeling 
Cross-Validation PPI targets and peptide complexes defy single-chain prediction. We deliver AlphaFold-Multimer, AI-driven docking, and MD-refined assemblies --- validated by Cryo-EM or X-ray --- to reveal binding interfaces, allosteric hotspots, and druggable cryptic sites.
Why Complex Structure Prediction Is the Critical Foundation
Structure-based drug design without a reliable 3D model is navigation without a map: seed-stage biotechs lack protein infrastructure to validate hits internally, and pharma teams pursuing membrane proteins or PPIs need models that survive experimental scrutiny. Our platform eliminates that uncertainty by generating AlphaFold predictions, refining them with microsecond-scale MD, scoring pockets with AI druggability classifiers, and validating critical regions through co-crystallization or Cryo-EM SPA --- delivering a model you can trust for virtual screening, docking, and medicinal chemistry decisions.
What Sets the Platform Apart
AI + Physics Hybrid
AlphaFold-Multimer and AI docking predict global architecture; MD resolves interface side-chain rotamers and captures induced-fit motions.
Interface-First Validation
Priority complexes advance directly into gene-to-protein production and experimental structure determination, delivering validated coordinates.
Druggability-Scored
Every interface is processed through binding pocket analysis: cryptic sites and allosteric hotspots are scored by ML classifiers trained on 500+ co-crystal structures.
Technology Suite
AI-Driven Complex Assembly & Interface Prediction
High-Accuracy Multimer Modeling with Uncertainty Quantification
Key Features:
- AlphaFold-Multimer v3 --- Predicts heteromeric and homomeric assemblies up to 2,000+ residues with ipTM/pAE confidence metrics; optimized for transient and stable complexes.
- AI Docking Refinement --- HADDOCK/RosettaDock hybrid protocols integrate AlphaFold predictions with experimental restraints (crosslinking, mutagenesis) to refine interface geometry.
- Disorder & Flexible Loop Annotation --- Flags intrinsically disordered regions and low-confidence interface loops for ab initio refinement or experimental determination.
Ideal For: PPI targets without experimental structures; antibody-antigen complexes; signaling cascades; membrane protein oligomers.
What We Offer:
Virtual biotechs access pharma-grade assemblies without building a structural biology department. Pharma teams receive parallel-path prediction: while internal crystallography battles a stubborn GPCR, we deliver an MD-refined assembly ready for docking and FEP within weeks.
Interface Dynamics & Cryptic Site Discovery
From Static Assembly to Dynamic Interaction Landscape
Key Features:
- All-Atom MD Interface Sampling --- GROMACS/AMBER simulations sample induced-fit motions, side-chain rotamer flips, and interface breathing, producing 50--200 conformers for ensemble docking.
- Cryptic Pocket Discovery at Interfaces --- Trajectory analysis identifies transient allosteric pockets and peptide-binding grooves invisible in static models.
- Hotspot Mapping --- Computational alanine scanning and ML classifiers score interface residues by binding energy contribution and druggability.
Ideal For: Transient PPIs; allosteric modulator programs; peptide-binding grooves; antibody paratope refinement.
What We Offer:
AlphaFold gives one rigid assembly; biology gives an ensemble. Our refinement generates representative conformers for virtual screening against the full interface landscape. For seed-stage biotechs, this finds non-obvious binding sites without cryo-EM budgets. For pharma, ensemble data feeds directly into FEP calculations to improve affinity correlation.
Experimental Validation (X-ray / Cryo-EM / NMR)
Closing the Computational-Experimental Loop
Key Features:
- Gene-to-Structure Integration --- Predicted assemblies guide construct design: subunit boundaries, linker engineering, and solubility tags are selected based on interface confidence.
- Co-crystal Soaking --- Models inform soaking conditions and cryoprotectant selection, reducing crystal optimization from months to weeks.
- Cryo-EM Model Building --- Predictions serve as initial models for single-particle analysis, accelerating map-to-model fitting for large complexes.
Ideal For: Programs requiring IND-grade structural evidence; membrane protein oligomers; large signaling assemblies; antibody-antigen complexes.
What We Offer:
When programs advance to lead optimization, experimental validation is required --- not a confidence score. Our structural biology team produces the protein complex, validates the assembly, and delivers PDB coordinates with electron density maps. Zero handoffs.
Platform Instrumentation
Core Instruments
| Instrument | Capability |
|---|---|
| NVIDIA DGX A100 | AlphaFold-Multimer inference + MD simulation at scale |
| NVIDIA RTX A6000 Cluster | Real-time MD visualization and ensemble analysis |
| GROMACS/AMBER HPC | Microsecond-scale all-atom MD; ensemble docking processing |
| Bruker AVANCE NEO 600 MHz | NMR validation of interface conformations |
| Rigaku XtaLAB Synergy | X-ray diffraction for co-crystal validation |
| Thermo Fisher Krios G4 | Cryo-EM single-particle analysis for complex validation |
Standardized Workflow
Project Workflow
A milestone-driven execution system from sequence to validated assembly.
01 Target Review
- Sequence analysis and domain annotation
- Stoichiometry and interaction partner assessment
- Deliverable: Target assessment report + construct proposal
02 AI Assembly
- AlphaFold-Multimer prediction with ipTM/pAE confidence mapping
- Model quality assessment and low-confidence region flagging
- Deliverable: Raw prediction + confidence map + quality report
03 MD Refinement
- All-atom MD simulation for interface refinement and pocket dynamics
- Ensemble clustering (50--200 conformers)
- Deliverable: Refined ensemble + trajectory analysis
04 Interface Score
- Pocket detection across all ensemble members
- ML druggability scoring and cryptic site ranking
- Selectivity indexing against human structural proteome
- Deliverable: Pocket prioritization report + druggability scores
05 Validation
- Gene-to-protein production for validation (optional)
- Co-crystal soaking or Cryo-EM SPA
- Model-to-experiment deviation analysis
- Deliverable: Validated assembly + experimental data + final report
Sample Requirements
- Target Sequences: Amino acid sequences of all subunits in FASTA format; UniProt IDs acceptable
- Stoichiometry: Known subunit ratio and interaction partners
- Prior Structural Data: Existing PDB entries or homology models of individual subunits
- Restraints: Crosslinking, mutagenesis, or FRET data (if available)
- Project Background: Target class, disease relevance, known challenges
Standard Deliverables
- AlphaFold-Multimer prediction with ipTM/pAE confidence coloring (PDB)
- MD-refined conformational ensemble (50--200 representative PDBs)
- Interface analysis report with hotspot scores and cryptic site maps
- Experimental validation data (if selected): X-ray or Cryo-EM coordinates
- Final technical report with quality metrics and SBDD recommendations
- Electronic data package (raw predictions, MD trajectories, analysis scripts)
Frequently Asked Questions
Case Study
GPR75: Multi-Method Determination of a Peptide Agonist Binding Mode on an Orphan GPCR
Goal: Determine where a peptide agonist binds the orphan Class-A receptor GPR75 — no prior structural data available — using an integrated computational-experimental workflow.
Key Data:
- Prediction: ColabFold complex prediction with PAE analysis identified a high-confidence peptide-receptor interface in the extracellular vestibule.
- Convergence: Cryo-EM difference density independently resolved density in the same region, converging with computational PAE on the same six-residue motif (VREFIW) without cross-method feedback.
- Deliverable: Validated binding mode with mapped interaction residues, ready for structure-guided medicinal chemistry.
Why it matters: Orphan GPCRs with no known binding site are among the hardest starting points in drug discovery. Standard approaches — blind docking or cryo-EM alone — each risk false positives. This case shows that running AI-driven complex prediction and cryo-EM difference density in parallel, then cross-validating their independent outputs, yields a single defensible binding mode without iterative back-and-forth. For discovery teams facing poorly characterized targets, this parallel framework reduces artifact risk and accelerates the path to structure-guided chemistry.
Figure 1. Cryo-EM difference density map of GPR75 revealing unassigned extra density in the extracellular vestibule.
Figure 2. ColabFold Predicted Aligned Error (PAE) matrix confirming a high-confidence interface between receptor chain A and peptide chain B.
Final validated binding mode: peptide (yellow) docked into the VREFIW motif of the GPR75 extracellular vestibule.
Our technical team responds within 24 hours. All inquiries protected under NDA.
