Humanized Framework
CDR Library Design
AI Affinity Scoring
Yeast Expression
ELISA + WB Validation
What Is a VHH Antibody?
VHH antibodies — also known as nanobodies or single-domain antibodies — are the variable domains of camelid heavy-chain-only immunoglobulins. At approximately 15 kDa, they are the smallest naturally occurring antigen-binding fragments, yet retain full binding specificity and affinity comparable to full IgG antibodies.
Their unique structure — a single immunoglobulin fold stabilised by a conserved disulfide bond and an extended, convex CDR3 loop — allows them to penetrate enzyme active sites, receptor pockets and cryptic epitopes that are entirely inaccessible to the flat paratopes of conventional IgG antibodies.
ICDMO's AI screening platform applies deep learning to explore the CDR sequence space of humanised VHH scaffolds at scale, identifying high-affinity binders in one week that would otherwise require months of library screening.
Antibody Format Comparison
Full IgGCannot reach cavity epitopes~150 kDa
FabLimited tissue penetration~50 kDa
scFvAggregation-prone linker~25 kDa
VHHDeep epitope access, stable~15 kDa
Size proportional to molecular weight
Key Characteristics of VHH Antibodies
1
Compact ~15 kDa Size
Smallest fully functional antibody fragment — accesses cryptic epitopes, enzyme active sites and receptor grooves sterically inaccessible to conventional IgG.
2
Thermal & Solvent Stability
Extraordinary resistance to high temperatures, detergents and extreme pH — retains activity under conditions that denature conventional antibodies.
3
Deep Tissue Penetration
Compact structure enables superior tumour penetration, blood-brain barrier crossing, and intracellular delivery for therapeutic applications.
4
Easy Engineering
Single domain simplifies formatting into bispecific, trispecific, CAR-T constructs, VHH-Fc fusions, and nanobody-drug conjugates.
5
Low Immunogenicity
Humanized frameworks with >90% sequence identity to human VH3 germline minimise immunogenic risk for therapeutic programmes.
6
Camelid Origin
Naturally occurring in llama, alpaca and camel IgG heavy-chain-only antibodies — diverse CDR3 loops enable unique binding modes.
AI Screening Platform
AI Workflow
1
CDR Mutagenesis
CDR1/2/3 regions undergo controlled random mutagenesis, generating a virtual library of millions of VHH sequence variants around the humanized framework scaffold.
2
Deep Learning Scoring
Deep learning models trained on VHH–antigen co-crystal structures score predicted binding energetics, shape complementarity and hydrogen-bond networks for every variant. Scores rank variants; they do not measure affinity.
3
Top Candidate Selection
Ranked candidate list filters for top 200–500 predicted binders, applying additional filters for solubility, aggregation propensity, and expression compatibility in yeast secretion systems.
Model Architecture
Transformer + Graph Neural Network
Training Dataset
>50,000 VHH–antigen structures
Prediction Output
ΔΔG, shape complementarity, H-bonds
Five-Step Service Workflow
1
Humanized Framework Design
Natural VHH sequences are back-grafted onto human VH3 germline frameworks. CDR loops are preserved while all framework residues achieving ≥90% human identity — maintaining function while minimising immunogenicity risk.
2
CDR Random Mutagenesis
CDR1, CDR2 and CDR3 regions undergo controlled random mutagenesis, generating a virtual library spanning millions of sequence variants around each humanized scaffold.
3
AI Deep Learning Screening
Deep learning models score every variant for predicted antigen binding, shape complementarity and structural stability. Top-ranked sequences are put forward for synthesis and testing.
4
Yeast Secretion Expression
Selected VHH candidates expressed using optimised Pichia pastoris or S. cerevisiae secretion systems. Secreted VHH purified by IMAC and confirmed by SDS-PAGE and SEC-HPLC.
5
ELISA & Western Blot Validation
Purified VHH antibodies validated by direct and competition ELISA (EC50, Kd estimation) and Western blot. Confirmed high-affinity binders delivered with full sequence data and validation report.
Applications
Therapeutic nanobodies against GPCRs, ion channels and enzyme targets
VHH-drug conjugates (NDC) for precise tumour targeting
Agonist / antagonist nanobodies for signalling pathway modulation
Bispecific and trispecific VHH-Fc therapeutic formats
Case Study Results
The following results were achieved across completed VHH AI screening projects. Each programme ran from sequence submission to confirmed validated binder in under two weeks.
Drug Discovery Target95% success
Hit identified in ≤1 week
GPCR Conformational VHH88% success
Active-state selective binder
Tumour Antigen Target92% success
Sub-nM Kd confirmed by SPR
Viral Surface Protein97% success
Neutralising nanobody identified
Platform Track Record
Projects Completed200+
Avg. Time to Hit<1 week
Validated Binders Delivered1,800+
Expression Success Rate99%+
Sub-nM Kd Rate>60%
Service Modules & Turnaround
Is This the Right Service for You?
Choose this when
Nanobody candidates from immunisation or display need triage before expression
A VHH must be humanised without losing stability
Multivalent or biparatopic constructs are being designed from VHH modules
Consider another service when
A conventional antibody is the format
Antibody Drug Design Only structural models are needed
Antibody Structure Prediction & Optimization What Goes Wrong, and How We Handle It
These are the failure modes that produce confident-looking results which do not survive the next experiment. Each one is designed against rather than discovered afterwards.
Panning enrichment used as the selection criterion
Candidates are ranked on developability and modelled behaviour as well as enrichment, since the most enriched clone is often not the best molecule.
Framework-2 hallmark residues germlined without compensation
Those positions are treated explicitly because they carry solubility in the absence of a light chain.
Multivalent constructs assembled without modelling reach
Linker sampling establishes whether both modules can engage simultaneously before constructs are built.
A Worked Example
Choosing twenty clones from several hundred
The situation
A display campaign produced a large sequence set and limited expression capacity.
What the work involved
Sequences were clustered by CDR3, germlines assigned, PCR artefacts flagged, leading clones modelled with CDR3 ensembles, and surfaces screened for aggregation-prone regions and chemical liabilities.
What came back
Nine genuine families, four carrying liabilities that would have surfaced later, and a final twenty constructs spanning the clean families with the reasoning recorded per clone.
Illustrative of the shape this work takes. Details are generalised, no client or target is identified, and scope and outcome vary with the system in question.
What Published Work Reports
The results below were published by independent groups and are cited to show what these computational approaches have achieved in the literature. They are not ICDMO project results, and outcomes vary substantially between targets, datasets and programmes. We do not promise comparable results for any particular project.
10- to 140-fold affinity improvements from computational design
Computational interface redesign produced a tenfold affinity improvement in cetuximab, reaching 52 pM, and a 140-fold improvement to 30 pM for an anti-lysozyme model antibody — improvements achieved by designing a small number of interface substitutions rather than by library selection.
Lippow, Wittrup & Tidor, Nature Biotechnology, 2007 ↗Interface models predict which substitutions help, with measurable accuracy
A 2024 study combining structural interface analysis with sequence-derived evolutionary information reported a predictive model for antibody–antigen interaction with an area under the curve of 0.83 and precision of 0.89 on held-out data, and experimentally confirmed a designed point mutation that improved affinity 2.5-fold to 2 nM. Typical computational maturation gains are modest per mutation, which is why a designed panel is measured rather than a single variant.
Experimental validation study, 2024 ↗What We Need From You
▸VHH sequence or sequences, with measured affinity, stability or expression data where available
▸Antigen sequence or structure, plus epitope information if mapping has been performed
▸Sequencing output from immunisation or display selection, where a repertoire is to be triaged
▸The design objective: affinity, specificity, stability, humanisation or format conversion
▸Intended application, since diagnostic, imaging and therapeutic uses impose different constraints
How the Analysis Is Performed
▸Repertoire analysis and clustering where sequencing data from selection campaigns is supplied
▸VHH structure prediction with dedicated CDR3 loop modelling and conformational ensemble generation
▸Framework analysis against camelid and human germlines, including the hallmark framework-2 positions
▸Antigen complex modelling and epitope-directed loop design where the target site is known
▸Sequence optimisation applying stability and solubility constraints alongside affinity objectives
▸Developability screening with aggregation-prone region detection on the modelled surface
What You Receive
▸Ranked VHH candidates or designed sequences with the rationale for each modification
▸Structural models including CDR3 ensembles and, where applicable, antigen complexes
▸Humanness assessment with humanisation designs and back-mutation recommendations
▸Developability report covering aggregation propensity, solubility, charge distribution and chemical liabilities
▸Prioritised construction and testing plan across the delivered panel
Interpretation and Limitations
Long CDR3 loops remain among the harder modelling problems, so designs are checked for robustness across an ensemble rather than a single predicted pose. Humanisation of the framework-2 hallmark residues can cost solubility and stability, which is why graded series are delivered for measurement rather than a single construct. Predicted affinity and aggregation values indicate risk and rank order, not outcome, and deliverables are research-use only.
Questions We Are Asked Before Scoping
Can you triage sequencing output from a selection campaign?
Yes. Clustering, germline analysis, structural modelling and developability screening across a repertoire is one of the most efficient uses of computation in nanobody programmes.
Will humanisation cost affinity or stability?
It can, particularly at the hallmark framework-2 positions. Designs are delivered as a graded series so the trade-off is measured across a small panel rather than guessed at.
Can you design a VHH against a new antigen from scratch?
De novo binder design is possible but low-yield and best treated as exploratory. Engineering a binder you already have from immunisation or display is considerably more likely to succeed.
Get a Custom Quote
Submit your target antigen and receive a project plan and quote within 24 hours.
Contact Us Online ConsultationStandard Deliverables
✓Humanized VHH sequence design report
✓AI-ranked candidate list + affinity scores
✓Purified protein (SDS-PAGE, SEC-HPLC)
✓ELISA binding data (EC50)
✓Western blot confirmation
✓Full project report (PDF)
Why ICDMO?
Models Trained on Public Structure Data
Design → ranked shortlist → partner validation
Confidence Metrics on Every Variant
Timeline Confirmed After Scoping
Response Within One Business Day