Overview
What this category covers
Protein engineering ranges from the routine to the genuinely experimental, and this category is explicit about which is which. Stability engineering is reliable and generally delivers. Enzyme engineering works well when computation designs the libraries and experiment supplies the feedback. De novo design and membrane protein design succeed at rates measured in percent, and we size panels accordingly rather than overselling them.
What we deliver
▸De novo binder, scaffold and assembly design with in silico filtering before expression
▸Thermostability and aggregation engineering with functional sites protected
▸AI-guided enzyme engineering with focused libraries and round-by-round model retraining
▸Membrane protein design and thermostabilisation in realistic bilayer environments
▸Construct recommendations that account for expression and purification reality
Who this is for
▸Industrial biotechnology teams improving enzyme performance
▸Groups needing a protein stabilised for manufacturing, formulation or structural study
▸Programmes exploring designed miniprotein binders as antibody alternatives
▸Structural biology projects blocked by an unstable membrane protein
How engagements in this category are run
Scoping starts with an honest statement of which regime the project sits in. Stability and enzyme engineering deliver reliably and are scoped as such; de novo and membrane protein design are numbers games and are scoped with panel sizes and expected success rates written into the proposal. Designs are filtered computationally before expression, and where a campaign requires several rounds we plan the feedback loop rather than treating the first panel as the deliverable.
Services
4 computational services in this category
K1
De Novo Protein Design
Design of entirely new proteins with defined 3D fold, binding function, or catalytic activity. Hallucination, inpainting, and motif-scaffolding approaches.
Learn MoreK2
Protein Stability & Thermostability Engineering
Computational engineering of proteins for improved thermal stability (Tm), pH stability, oxidative resistance, and shelf life. Point mutation scanning, consensus sequence analysis, and direc…
Learn MoreK3
Enzyme Design & Directed Evolution (AI-Guided)
ML fitness landscape modeling for enzyme variants. Activity, enantioselectivity, and substrate scope optimization. Saturation mutagenesis library prioritization, adaptive laboratory evolutio…
Learn MoreK6
Membrane Protein Design
Computational design and engineering of transmembrane proteins (ion channels, GPCRs, transporters). TM helix packing optimization, lipid-facing residue design, and functional gating mechanis…
Learn MoreIn Practice
What a project in this category looks like
Three shapes we are asked for most often. Scope is always set after a feasibility review, so timelines below are typical rather than quoted.
Raising the melting temperature of a process enzyme
The situation
An enzyme performs well but unfolds at process temperature, limiting throughput and shelf life.
What we do
Computational saturation mutagenesis with folding free energy prediction, consensus and evolutionary analysis, functional site protection, and combination design modelled rather than summed.
What you receive
Ranked single mutations and modelled combination variants with predicted stability gains, functional risk flags, and a testing order where early results inform later rounds.
Typical timeline: 3–6 weeks
A multi-round enzyme engineering campaign
The situation
Activity or selectivity must improve substantially and random mutagenesis has plateaued.
What we do
Mechanistic and active site analysis, position prioritisation from structure and evolutionary coupling, fitness modelling on whatever variant data exists, focused library design sized to your screening throughput, and model retraining between rounds.
What you receive
A library specification or designed variant set each round, structural analysis of the leading designs, and a campaign plan stating what data drives the next round.
Typical timeline: 4–8 weeks per round, 2–5 rounds
Designed miniprotein binders as an antibody alternative
The situation
A small, stable, cheaply produced binder is wanted for imaging, delivery or reagent use.
What we do
Backbone generation against the target surface, sequence design, self-consistency filtering, interface and stability assessment, and expression-risk screening before anything is made.
What you receive
A filtered panel with per-design metrics, interface analysis, construct suggestions, and a realistic statement of expected success rates.
Typical timeline: 6–10 weeks
Questions
Asked before we scope this work
Which of these is reliable and which is exploratory?
Stability engineering delivers dependably. Enzyme engineering works well when computation designs the libraries and experiment supplies feedback. De novo binder design and membrane protein design succeed at rates measured in percent, and we scope them that way.
How much variant data do we need?
None to start. A few dozen measured variants improve the models noticeably; several hundred allow genuinely predictive modelling. Data returned between rounds is what makes later rounds better.
Will stabilisation cost activity?
It can, since flexibility often matters for function. Functional sites are protected in the design and every variant is delivered with a functional risk flag so activity is measured alongside stability.
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.
Machine-learning-guided evolution raised a reaction yield from 12% to 93%
An active-learning workflow optimising five interacting active-site residues improved the yield of a desired non-native cyclopropanation product from 12% to 93% in three rounds of laboratory experimentation. The design principle is the one we apply: computation chooses which small set of variants to make, and measured results retrain the model between rounds.
Nature Communications, 2025 ↗1.6- to 42-fold activity improvements from model-selected enzyme variants
A machine-learning-guided enzyme engineering campaign evaluated 1,217 variants across 10,953 reactions to build predictive models for amide synthetases. Model-selected variants showed 1.6- to 42-fold improved activity relative to the parent enzyme across nine pharmaceutical targets, illustrating both the scale of data that makes models predictive and the spread of outcomes that remains.
Nature Communications, 2024 ↗De novo binder design succeeds at rates that vary enormously by target
Recent design methods report nanomolar binder hit rates between roughly 17% and 82% across evaluated targets, while independent evaluations of earlier pipelines on other targets report designs failing on expression, non-specific binding or undetectable affinity. The honest reading is that de novo binder design now works, that success is strongly target-dependent, and that panels must be sized accordingly rather than on a best-case figure.
Comparative design studies, bioRxiv, 2025 ↗More
Other service categories
Target Identification & Validation5Virtual Screening (Small Molecule)3Small Molecule Design & Optimization2ADMET & Drug-Likeness Prediction4Molecular Dynamics & Simulation3AIDD — Small Molecule Platform2Structure-Based Drug Design (SBDD)3Bioinformatics & Multi-Omics Analysis11Antibody Design & Engineering (AI)5Peptide & Cyclic Peptide Drug Design4ADC & Bioconjugate Design2Compound & Biological Library Design3
Bring us the scientific question, not a service code
Share your target, dataset, sequences or objective. We will recommend a scoped computational plan, the expected deliverables, a timeline and the validation options.
Request a Feasibility Review