Research overview
Nature — Protein engineering
Current protein-engineering research.
Open Nature topic ↗Diversity → design → mutations → application → manufacturing
Design proteins for useful chemistry.
Protein Design Space is an independent scientific webspace for enzyme discovery, protein engineering, biocatalysis, green chemistry, computational design, patent intelligence and scalable biomanufacturing.
PDS
General workflow
From biological diversity to industrial application
A practical protein-engineering workflow connects biological diversity, sequence and structure analysis, variant design, expression, screening, process development and application validation. The aim is to move systematically from discovery to an industrially relevant solution while considering performance, scalability, sustainability and intellectual-property space.
01Diversity & discoveryGenomes, metagenomes, homologues, phylogeny and functional annotation. Search beyond familiar scaffolds before mutating the same protein repeatedly.
02Sequence, structure & mechanismConnect sequence motifs, predicted or experimental structures, active sites, reaction mechanism, channels and conformational constraints.
03Mutation & library designRational substitutions, directed evolution, saturation libraries, second-shell positions, stability networks, interfaces, tunnels and active-learning selection.
04Construct & host selectionChoose expression architecture and host around folding, cofactor needs, soluble expression, secretion, enzyme titre and production economics.
05Expression, production & screeningProduce variants efficiently, purify or use whole-cell/lysate formats, then screen activity, selectivity, stability, solvent tolerance and substrate loading.
06Reaction & catalyst engineeringOptimise pH, temperature, cosolvent, cofactor, donor/acceptor, substrate feed, enzyme loading, immobilisation and catalyst recycling.
07Application & process validationDemonstrate target conversion, product quality, isolation and analytics at proof-of-concept scale, then benchmark gram, kilo and manufacturing-relevant performance.
08TEA, LCA & patent intelligenceTrack cost-of-use, yield, productivity, PMI, E-factor, energy and waste alongside patent claims, sequence scope, legal status and design-around opportunities.
Design principle — “You get what you screen for.”
In protein engineering, an evolved protein improves the traits that the assay or selection directly measures and rewards. Design the screen around the property you actually want.
Best protein engineering strategy for your business needs
Choose the engineering mode that fits the problem
The right programme depends on available time and resources, how crowded the technical or IP space is, and whether you already have a workable starting scaffold.
Resources & time: fast / limited → deep / well resourced
Design context: known / constrained → novel / open
Low resources · less time
Fast-frugal engineering
When speed and experimental capacity are limited.
- Start from the best validated scaffold.
- Use a small MSA, consensus and one good structure.
- Test a focused set of 5–20 high-information variants.
- Prioritise clear stability, pocket or tunnel hypotheses.
Mode: Rational + semi-rational
More resources · more time
Full-stack engineering
When the programme can support deeper learning and multiple rounds.
- Mine broad diversity and alternative scaffolds.
- Generate kinetics, stability, selectivity and expression data.
- Run focused libraries followed by iterative evolution.
- Bring process and manufacturability into the design cycle early.
Mode: Semi-rational + directed evolution
Competitive · patent-protected area
IP-aware differentiation
When obvious sequence and mutation space is already crowded.
- Map claims, patented sequences and mutation positions first.
- Explore distant homologues and alternative enzyme families.
- Engineer tunnels, second-shell sites, interfaces and process format.
- Differentiate by substrate scope, robustness, reuse or manufacturing route.
Mode: Rational + semi-rational + design-around
Starting from scratch · novel molecule
Frontier discovery
When there is no validated enzyme or starting scaffold.
- Define the reaction, mechanism and assay first.
- Search broadly across enzyme classes and metagenomic diversity.
- Screen for weak function before optimising performance.
- Move from broad discovery into focused engineering after the first hit.
Mode: Random / “irrational” → semi-rational → rational
Engineering modes: Rational = hypothesis-driven mutations · Semi-rational = focused libraries around informed hotspots · Random / “irrational” = broad mutagenesis where the screen identifies winners.
Open scientific toolbox
Resources organised by the protein-design workflow
A compact, task-based set of resources for discovery, structure, mutation design, visualisation and IP.
Foundations & key reading
Protein engineering concepts from specialist and primary sources
Biocatalysis scale-up
Integrating protein engineering into biocatalytic process scale-up
Protein engineering for process scale-up.
Open DOI ↗Directed evolution
Protein Design by Directed Evolution
Directed-evolution principles.
Open DOI ↗Semi-rational design
Beyond directed evolution
Semi-rational design strategies.
Open DOI ↗Enzyme engineering & biocatalysis
Kazlauskas Lab
Enzyme engineering and biocatalysis.
Open resource ↗Sequence discovery & protein properties
Find diversity before designing mutations
Find homologues and sequence patterns.
Sequence search
NCBI BLAST
Homologue sequence search.
Open BLAST ↗Protein annotation
UniProt
Sequence and functional annotation.
Open UniProt ↗Sequence-property profiling
ExPASy ProtScale
Sequence-property profiles.
Open ProtScale ↗Multiple sequence alignment
Clustal Omega
Multiple-sequence alignment.
Open Clustal Omega ↗Multiple sequence alignment
MAFFT
Fast large-family alignment.
Open MAFFT ↗Multiple sequence alignment
MUSCLE
Alignment for conservation analysis.
Open MUSCLE ↗Multiple sequence alignment
T-Coffee
Consistency-based alignment.
Open T-Coffee ↗Structure prediction & structural references
Move from sequence to three-dimensional hypotheses
Retrieve or build 3D models.
Predicted structures
AlphaFold Protein Structure Database
Predicted protein structures.
Open AlphaFold DB ↗Homology modelling
SWISS-MODEL
Homology structure modelling.
Open SWISS-MODEL ↗Experimental structures
RCSB Protein Data Bank
Experimental protein structures.
Open RCSB PDB ↗Structure prediction & design
RosettaCommons
Protein modelling and design.
Open Rosetta ↗Structure prediction
I-TASSER
Structure and function prediction.
Open I-TASSER ↗Remote homology / templates
HHpred
Remote homology and template search.
Open HHpred ↗Mutation design & stability engineering
Prioritise positions and mutations
Choose hotspots and stabilising variants.
Mutation hotspot discovery
HotSpot Wizard
Mutagenesis hotspot selection.
Open HotSpot Wizard ↗Stability engineering
FireProt
Stability mutation design.
Open FireProt ↗Tunnels, channels & access pathways
Engineer how substrates and products reach the active site
Analyse substrate and product access.
Tunnel analysis
CAVER
Tunnel and access-path analysis.
Open CAVER ↗Channels & pores
ChannelsDB 2.0 — Methods
Channel and tunnel analysis.
Open ChannelsDB 2.0 ↗Molecular visualisation & dynamics
Inspect, compare and communicate structural hypotheses
Inspect structures and trajectories.
Molecular graphics
PyMOL
Protein structure visualisation.
Open PyMOL ↗Visualisation & MD analysis
VMD
Molecular-dynamics visualisation.
Open VMD ↗Patent landscape & biological-sequence IP
Connect design space with intellectual property space
Map claims and patented sequences.
Patent & scholarly landscape
Lens.org
Patent and scholarly search.
Open Lens ↗Patent sequence search
Lens PatSeq Finder
Patent-disclosed protein sequences.
Open PatSeq Finder ↗Green chemistry examples
One design logic, many chemistries
Examples below are shown as technology case studies.
Sitagliptin
ω-Transaminase catalysis is a classic example of engineering an enzyme for a demanding chiral-amine transformation.
200 g/Lketone loading in
>99.9% eereported selectivity
90–98%reported yield range
PMI 23vs 30 chemical in
Montelukast intermediate
A ketoreductase / alcohol-dehydrogenase route demonstrates how protein engineering can replace a sensitive stoichiometric chiral-reduction reagent.
100 g/Lsubstrate loading
>99.9% eecrude product
45°Creported biocatalytic condition
PMI 18–34vs 52 chemical
Pregabalin
The lipase-enabled pregabalin manufacturing as an example of improved yield, lower process mass intensity and lower energy demand.
44–45%reported biocatalytic yield
E-factor 17vs 86 chemical
PMI 11.68vs 57.9 chemical
21 MJ/kgvs 118 MJ/kg energy
Beyond APIs
The same workflow applies to industrial enzymes, food biocatalysis, specialty chemicals, environmental enzymes, alternative proteins and precision-fermented products: discover diversity, engineer function, build the expression system, validate application and scale.
Discoverydiversity → candidates
Translationactivity → process value
Protein Design Space is intentionally application-agnostic.
Broader process redesign
Green chemistry is larger than biocatalysis
The specific pharmaceutical process-redesign examples. These are useful benchmarks for waste reduction even when the redesigned route is not necessarily enzyme-based.
| API | Example sponsor | Reported waste decrease | How it informs protein design |
|---|---|---|---|
| Sertraline HCl | Pfizer / Zoloft | 92% | Shows the scale of process simplification worth targeting. |
| Sildenafil citrate | Pfizer / Viagra | 93% | Benchmark for solvent, reagent and unit-operation reduction. |
| Celecoxib | Pfizer / Celebrex | 69% | Reminds enzyme projects to compare against redesigned chemistry, not legacy chemistry only. |
| Pregabalin | Pfizer / Lyrica | 80% | Connects catalytic selectivity to large reductions in process waste. |
| Quinapril HCl | Pfizer / Accupril | 80% | Useful benchmark for route-level sustainability. |
| Sitagliptin | Merck / Januvia | 80% | Illustrates how enzyme engineering can become manufacturing innovation. |
| Paclitaxel | BMS / Taxol | >90% | Highlights the value of route redesign for complex molecules. |
| Nevirapine | Medicines for All Institute | 93% | Demonstrates the importance of whole-route optimisation. |
| Ganciclovir | Roche Colorado Corp. | 89% | Another benchmark for material-efficiency improvement. |
For beginners · AI-enabled protein engineering
Use AI to narrow the search — experiments still decide
A simple workflow for moving from large sequence and mutation spaces to a small, testable set.
Goal → Define the property you want to improve.
Diversity → Collect homologues or metagenomic candidates.
Consensus → Align sequences and identify conserved or variable sites.
Structure → Map active sites, pockets, interfaces and tunnels.
Hotspots → Prioritise positions most worth mutating.
AI scoring → Rank plausible substitutions before experiments.
AI design → Generate a focused set of plausible variants.
Experiment → Test a small, information-rich library.
Iterate → Use measured results to guide the next round.
Beginner mental model:
Evolutionshows what nature conserves or varies.
Structureshows where residues sit and interact.
AIhelps rank or generate plausible choices.
Experimentsdecide what actually works.
Beginner worked example
How could you choose one mutation from hundreds of possibilities?
This is a simple hypothetical example to show the logic. Imagine your enzyme has Methionine at position 148 (M148), and you want to know whether this position is worth engineering.
1 · Look at diversity
You collect about 500 related sequences and align them.
2 · Check consensus
At position 148 you find: Leu 61%, Ile 22%, Met 8%, Val 6%. Your protein's Met is relatively uncommon.
3 · Check the structure
M148 is not catalytic, but sits near a hydrophobic pocket and does not make an essential interaction.
4 · Ask AI / design tools
A sequence model and a structure-aware design tool both rank Leu and Ile as plausible substitutions.
5 · Test, do not assume
You build M148L and M148I, then measure activity, stability and expression against the original enzyme.
What did AI actually do?
It did not prove that M148L would be better. It helped reduce a huge mutation space to a small, testable set supported by evolution + structure + model predictions. The experiment still decides whether the variant is useful.
For beginners: if any part of the workflow is confusing, or you want help mapping your first protein-engineering project, connect with me.
Founder profile
Dr Vivek Srivastava
Industrial Biotechnology & Biomanufacturing Scientist
PhD molecular and cellular biologist with 19+ years of industrial biotechnology R&D experience spanning industrial enzymes, protein engineering, recombinant proteins and biologics, precision fermentation, downstream processing, analytics, scale-up and commercial translation.
19+years industrial biotechnology R&D
5international PCT patent families in enzyme engineering
1–100 Lpilot-scale process experience
Selected technical impact
Discovery to manufacturing
Industrial enzyme engineering
Protein and enzyme variant discovery, rational design, directed evolution, screening and global application programs.
Precision biomanufacturing
Fermentation, downstream processing, analytics, process troubleshooting and manufacturability.
Alternative proteins
Built an integrated 1–100 L mycoprotein pilot platform and developed a commercial-scale manufacturing concept.
Biologics & recombinant proteins
Multidisciplinary development across therapeutic proteins, peptides, vaccines, antibodies and microbial enzyme programs.
Design principles
What makes an enzyme industrially useful?
Catalytic activity alone is not enough. A useful industrial enzyme must combine performance, robustness, manufacturability and process compatibility.
catalytic activity
selectivity
broad substrate scope
thermostability
pH stability
protease resistance
solvent tolerance
substrate loading
cofactor economy
expression titre
reusability
downstream compatibility
PMI / E-factor
Patent-aware engineering
Map claims before building libraries
A strong protein-design programme can combine sequence and structure analysis with patent intelligence: identify claimed sequence families and mutation positions, then explore genuinely distinct scaffolds, underused structural regions and alternative process architectures.
Contact
Scientific discussion & collaboration
Protein engineering, enzyme discovery, biocatalysis, green chemistry, patent landscapes, technical reviews and industrial biotechnology.