This is a wish list for protein folding and engineering.

Wishlist
TODO
Other interesting targets
Structural protein design with machine learning
Protein Engineering Cheatsheet
Common protein engineering design mistakes
Redesigning an existing protein
Protein-Engineering Amino-Acid Motifs Cheat Sheet
Talks, videos, and transcripts
Cloud platforms for protein design
References

Wishlist

This wishlist contains some speculation and brain storming and shouldn't be considered completely viable for now.

Given a 3d shape (of some nanostructure), produce a protein's amino acid sequence that will consistently create that shape. (done as of 2023?)
Control over protein functional properties, such as catalytic domains and sites, as well as designing specific confirmational changes and control over conformation changes.
DNA data storage: faster polymerases
Proteins that make molecular display techniques easier (simplifying lab bench protocols) -- like mRNA display and ribosome display; easier molecular display would be very valuable for projects using directed evolution techniques.
Better protein-based nanopores for DNA sequencing, amino acid sequencing, and protein sensing.
Human-controlled DNA polymerase synthesis activity (choose each nucleotide), or an instrumented ribosome to control protein production regardless of mRNA content
Molecular protein lego: connect multiple legos together to build large-scale protein structures. This is generally useful for modeling and nanostructures. Binding by DNA addresses or other high affinity ligand specific techniques, for a stable toolbox of known protein structures and shapes and building up larger structures from small parts.
Protein mechanical logic: protein structures that have internal logic and state, based on mechanical motion or other catalytic reactions and interactions.
Generalized, fully-programmable ?molecular nanotechnology: programmable nanomachines and nanofactories that can produce other nanostructures to exact specifications, without uncertainty regarding protein folding.

TODO

What were those long-tube protein molecular-chemistry factories called? These were non-ribosomal peptide synthetases or NRPS. They are apparently natural, and they have multiple points of interest inside the tube that modify a molecule as it progresses along the protein.
more Baker lab references

Design of a hyperstable 60-subunit protein icosahedron

Other interesting targets

gene editing proteins (see gene editing for various effectors or components that could be used in protein engineering)
enzymes for ?DNA synthesis
molecular recording (like in vivo DNA-based recording devices, for debugging or otherwise, lineage tracing techniques, DNA ticker tape memory, see "Of toasters and molecular ticker tapes" by Kording)
protein binding affinity stuff (protein-protein interaction)
catalytic activity, enhancement of catalysis or reduction of catalysis
synthetic metabolisms
biosensors

Structural protein design with machine learning

Well, it's probably time to update this page... lots of recent progress in machine learning for protein design.

PoET: A generative model of protein families as sequences-of-sequences

PGR: A Graph Repository of Protein 3D-Structures

Protein structure generation via folding diffusion

Protein bioelectronics: a review of what we do and do not know

Driving current through single organic molecules

Unlocking de novo antibody design with generative artificial intelligence (2023)

Protein Engineering Cheatsheet

This might get a little sloppy.

Core Engineering Strategies (Mechanistic Foundations)

Strategy	Mechanistic Basis	Typical Applications	Key Design Levers
Coiled‑coil / Leucine‑zipper dimerization	Helical packing of hydrophobic residues; electrostatic surface charge	Stabilizing enzyme activity, scaffold assembly, synthetic signaling	Length of helical repeat, hydrophobic‑index, charge pattern. Stability is tuned with core-packing (a/d layers), electrostatic e/g pairs and helix length.
Site‑Directed Mutagenesis	Precise amino‑acid substitution at a defined position	Changing substrate specificity, thermostability, altering protease cleavage sites via targeted mutations at recognition motifs	Alanine scanning or saturation mutagenesis at catalytic or interface residues for functional mapping
Rational Active‑Site Redesign	Structural knowledge (X‑ray, cryo‑EM) + computational modeling	Enhancing catalysis, altering reaction mechanism	Catalytic triad geometry in serine-protease-like enzymes (many enzymes such as lyases, isomerases, radical SAM lack a triad), transition‑state stabilization via electrostatic pre-organisation or proton-shuttling residues, electrostatic complementarity
Fusion Protein Construction	Covalent linkage of two domains via a linker	Chimeric enzymes, tagging, biosensors	Linker length & flexibility (Gly‑Ser repeats vs. α‑helical), domain orientation
Affinity Tagging	Short peptide that binds a resin or antibody	Rapid purification, immobilization	His‑tag, GST, MBP, FLAG, Strep‑II
Protease‑Cleavable Tags	Tags containing a specific amino acid sequence recognized and cleaved by an added exogenous enzyme	Removing tags to yield near‑native proteins	TEV, PreScission, thrombin recognition sites
Self‑Cleaving Tags	Autocatalytic domains (e.g., inteins) that cleave via nucleophilic attack by the N-extein side-chain or upon induction (e.g., pH, thiols) without exogenous protease	Producing native‑length proteins; simplifies purification steps	Inteins (e.g., Ssp DnaB), N‑pro, 2A peptides
PTM Engineering	Introducing or removing phosphorylation, glycosylation, SUMOylation sites	Controlling activity, stability, localization	Adding consensus motifs, mutating serine/threonine/tyrosine, N‑glycosylation sequons (N‑X‑S/T)
PPI Modulation	Short peptide motifs (SH3, PDZ, WW) or nanobodies	Rewiring signaling, synthetic scaffolds	Interface grafting, hotspot mutation, nanobody epitope mapping
Flexible Linker Insertion	Gly‑Ser repeats or other flexible sequences (e.g., (Gly-Gly-Ser-Gly-Ser)n safer in mammalian cells to avoid O-glycosylation)	Reducing steric hindrance, allowing domain movement	Linker length, composition (Gly, Ser, Ala)
Unstructured Region Deletion	Removing intrinsically disordered tails or loops	Improving solubility, reducing proteolysis	Identification of low‑complexity sequences via sequence-based disorder predictors (e.g., IUPred, MetaDisorder)
Degron / Protein‑Degradation Tags	Short sequences recognized by ubiquitin‑proteasome or autophagy	Controlling turnover, studying dynamics	PEST sequences, N‑end rule degrons, auxin‑inducible degron (AID); note that AID is a plant degron; in non-plant cells an F-box protein (TIR1) must be co-expressed, otherwise the tag is inert.
Optogenetic Modules	Light‑responsive domains (LOV, CRY2, PhyB) fused to a target	Spatiotemporal control of activity	Chromophore pocket mutations, photocycle kinetics, helix rotation
CRISPR‑Cas Fusion	Cas9 or dCas9 (deactivated Cas9) fused to effectors (activators, repressors, epigenetic modifiers) via flexible linkers	Gene regulation, epigenome editing	Linker optimization, nuclear localization signals (effectors appended without domain swapping that disrupts HNH/RuvC folds)
Nanobody / VHH Fusion	Single‑domain antibody fused to enzyme or reporter	Targeted delivery, stabilization	Framework mutations, dimerization interfaces
Aptamer‑Based Tethering	Nucleic‑acid aptamer that binds a protein to bring it into proximity	Conditional activation, proximity labeling	Aptamer selection (SELEX), binding affinity tuning
Synthetic Scaffolds (SpyTag/SpyCatcher)	Isopeptide bond formed between Asp side-chain and Lys ε-NH₂ within the SpyCatcher domain	Modular assemblies, enzyme cascades	SpyTag/SpyCatcher pairing, spacer length
Protein‑DNA Fusion	DNA‑binding domain fused to an enzyme	Locus‑specific regulation	DNA‑binding motif design, linker orientation
Molecular Glue / PROTACs	Small molecule or peptide that links target to an E3 ligase	Targeted protein degradation	Ligand selection, linker length, binding affinity
Allosteric Modulator Domains	Domain that binds a small molecule to control activity	Chemical control of function	Binding pocket design, conformational coupling
Synthetic Riboswitch‑Controlled Expression	Riboswitch embedded in mRNA to modulate translation	Metabolic control, feedback loops	Aptamer‑to‑transformation, ligand binding site optimization
Humanization	Grafts non-human CDRs onto a human framework; subsequently optimizes to closest germ-line identity while preserving CDR backbone conformation, removing immunogenic patches without altering paratope geometry	Reducing immunogenicity of therapeutic antibodies, improving compatibility with human immune system	Framework residue selection, CDR backbone preservation, patch removal
Scaffold Repurposing	Grafts catalytic or binding domains onto a stable, expressible protein skeleton, using secondary‑structure elements with geometric compatibility (RMSD, loop length, orientation) as attachment points and requiring core optimisation to preserve the fold while introducing new activity	Creating engineered enzymes, chimeric receptors, synthetic scaffolds	Attachment sites by geometric compatibility, domain orientation, scaffold stability
Sub‑Cellular Targeting Motifs	Appends short, evolutionarily honed peptide codes—NLS, MLS, myristoylation, palmitoylation, ER‑retention, PTS1, mitochondrial TOM/TIM signals—to steer nascent chains through membrane translocons or vesicular trafficking pathways	Targeting proteins to nucleus, mitochondria, plasma membrane, secretory pathway	Motif placement, context‑dependent processing, signal peptide design
Kinase‑Based Trafficking Tags	Embeds phosphorylation‑dependent PDZ or 14‑3‑3 interaction motifs that couple protein movement to local kinase gradients, delivering cargo to presynaptic or postsynaptic densities	Synaptic protein targeting, signal‑dependent trafficking	Phosphorylation sites, interaction domain compatibility, spatial regulation
Ubiquitin Tagging for Localization	Installs mono‑ubiquitin or K63 chains to direct localization/signaling (K63 less efficient than K48 at initiating proteasomal degradation but still processible) while exploiting ubiquitin‑binding domain networks	Targeting proteins to specific cellular compartments, signaling hubs	Ubiquitin‑binding domain selection, linkage type, chain length
Glycosylation as Trafficking Passport	Positions N‑glycan sequons in surface loops to recruit calnexin/calreticulin quality‑control or Golgi‑lectin sorting receptors, influencing ER exit, apical vs. basolateral delivery, or lysosomal routing	Controlling protein trafficking, secretion, membrane localization	Sequon placement, loop accessibility, glycan processing pathways
Azobenzene Photocontrol	Site‑specifically incorporates azobenzene unnatural amino acids whose trans–cis isomerization toggles the distance/orientation between attachment points on the azobenzene moiety linked to side chains, modulating effective geometry between hosting residues to induce conformational changes, conferring reversible light‑switchable activity	Light‑controlled enzyme activity, conformational switches	Incorporation site, isomerization kinetics, structural impact
Unnatural Amino Acids (UAAs) as Bio‑Orthogonal Handles	Uses orthogonal tRNA/synthetase pairs to introduce p‑azido‑phenylalanine, alkynyl‑lysine, or other reactive UAA, enabling click chemistry, photocrosslinking, or post‑translational modifications impossible within the canonical set	Site‑specific labeling, cross‑linking, chemical biology probes	tRNA/tRNA‑synthetase specificity, metabolic stability, incorporation efficiency

Protein Classes – Designability vs. Expressibility

Protein Class	Designability (Structural/Computational)	Expressibility/Foldability (Cellular Context)	Typical Success Rationale (Mechanistic Levers)	Representative Examples
GPCRs	High – many crystal structures, AlphaFold predictions	Medium – need membrane environment, chaperones	Thermostabilizing point mutations, toggle-switch engineering (NPxxY forms the toggle-switch with Tyr-7.53), ionic-lock-stabilising substitutions	β2‑adrenergic receptor thermostabilization; biased agonist design
Ion Channels	High – pore architecture known	Medium – require lipid bilayers, proper gating	Selectivity filter size/charge, S4 voltage-sensor gating Arg residues, pore helix dipoles	Voltage‑independent channels; fluorescent voltage sensors
Enzymes	High – active‑site geometry often dictated by a few residues	Medium – some enzymes need cofactors or chaperones	Catalytic triad geometry, substrate‑anchoring subsites, transition‑state stabilization	High‑fidelity polymerase; proteases with novel cleavage sites
Antibodies / Nanobodies	High – phage/yeast display libraries	Medium – expression of full IgG can be challenging	CDR-H3 conformation, framework stability, Fc‑glycosylation	Bispecific antibodies; intracellular nanobodies
Optogenetic Proteins	High – modular domains	Medium – chromophore incorporation	Chromophore pocket mutations, photocycle kinetics, helix movements	Red‑shifted opsins; faster‑acting LOV domains
Receptor Tyrosine Kinases	Medium – dimerization and autophosphorylation complex	Medium – require extracellular ligand binding	Domain swapping, interface grafting, synthetic ligand activation	EGFR chimeric receptors
Transcription Factors	Medium – DNA‑binding specificity can be altered	Medium – expression stability variable	Base‑specific readout, DNA‑binding domain design (e.g., helix-turn-helix in some families), linker length	dCas9‑VPR activators; synthetic zinc‑finger TFs
Structural Proteins	Low – repetitive, β‑rich scaffolds	Low – hard to express recombinantly	Core packing hydrophobic residues	Engineered collagen scaffolds
Membrane Transporters	Medium – pore architecture known	Low – lipid‑dependent folding, assay difficulty	Substrate‑binding cavity mutagenesis, gating domain engineering (e.g., helical tilts)	Aquaporin selectivity tuning
Viral Coat Proteins	Medium – capsid symmetry known	Low – assembly fidelity critical	Surface loop modifications while preserving quasi-equivalent contacts	AAV capsid engineering; VLP display
Complex Multi‑Domain Enzymes	Low – inter‑domain interactions tight	Medium – modular assembly possible	Domain swapping, linker redesign, active‑site complementation	Engineered PKS for novel metabolites
Highly Glycosylated Proteins	Low – glycoform control difficult	Low – folding depends on glycosylation	Sequon engineering (glycoform control requires cell-line engineering such as glyco-knockouts)	Glycosylation‑site modulation
Humanized Antibodies	High – framework grafting is computationally tractable	Medium – expression of full IgG can be challenging	Framework residue selection, CDR backbone preservation, immunogenic patch removal	Therapeutic antibodies with reduced immunogenicity
Scaffold‑Based Enzymes	High – repeat or domain scaffolds are designable	Medium – some scaffolds require chaperones	Attachment sites by minimal perturbation of catalytic Cα positions, domain orientation, scaffold stability	DARPins, repeat proteins, synthetic scaffolds
Azobenzene‑Controlled Proteins	Medium – incorporation site selection	Medium – photo‑isomerization may affect folding	Site‑specific UAA incorporation, structural impact of isomerization	Light‑switchable enzymes, conformational switches
Ubiquitin‑Tagged Proteins	Medium – degron design is tractable	Medium – need to avoid proteolysis	Ubiquitin‑binding domain selection, linkage type, chain length	Targeted localization to endosomes or DNA‑damage foci

Key Design Levers for Each Class

Class	Primary Levers (Mechanistic)	Secondary Levers
GPCRs	Toggle-switch motifs (NPxxY), ionic lock, ligand‑binding pocket shape	G‑protein coupling interface, β‑arrestin recruitment sites
Ion Channels	Selectivity filter residues, S4 voltage-sensor gating Arg residues, pore helix dipoles	Allosteric modulatory sites (e.g., calmodulin binding)
Enzymes	Catalytic triad geometry, substrate‑anchoring subsites, electrostatic field, transition‑state stabilization	Loop flexibility, metal‑binding sites, cofactor interactions
Antibodies	CDR H3 conformation, framework mutations, FcγR binding motifs	Glycosylation sites, hinge flexibility
Optogenetic Proteins	Chromophore pocket residues (Schiff-base network in retinal-binding rhodopsins), photocycle‑determining residues	Helix rotation, light‑absorption wavelength tuning
RTKs	Dimerization arm rigidity, ligand‑binding domain modularity, kinase activation loop	Intracellular adaptor binding sites
Transcription Factors	DNA‑binding helix‑turn‑helix motif (in HTH families), base‑specific hydrogen bonds, linker length	Recruitment of co‑activators/repressors
Structural Proteins	Core packing hydrophobic residues, β‑strand register	Surface charge distribution
Transporters	Pore lining residues, gating domain conformational changes, substrate‑binding cavity	Lipid‑protein interaction sites
Viral Coat Proteins	Surface loop residues, inter‑subunit interface residues, capsid symmetry	Capsid maturation signals
Multi‑Domain Enzymes	Domain interface complementarity, linker flexibility, active‑site alignment	Allosteric regulation by small molecules
Highly Glycosylated Proteins	Glycosylation sequon placement, peptide backbone flexibility	Enzyme‑mediated glycan trimming sites
Humanized Antibodies	Framework residue selection, CDR backbone preservation, immunogenic patch removal	CDR length optimization, affinity maturation
Scaffold‑Based Enzymes	Attachment sites by minimal perturbation of catalytic Cα positions, domain orientation, scaffold stability	Loop grafting, interface redesign
Azobenzene‑Controlled Proteins	Incorporation site, structural impact of isomerization	Photocycle kinetics, light‑absorption tuning
Ubiquitin‑Tagged Proteins	Ubiquitin‑binding domain selection, linkage type, chain length	Degron placement, proteasome interaction motifs

Missing / Under‑Represented Protein Classes (Additions)

Class	Why It Matters	Representative Design Strategies
Aptamers (RNA/DNA)	High‑affinity, programmable nucleic‑acid ligands; can be fused to proteins or used as biosensors	SELEX, computational aptamer design, in‑vitro selection
Repeat Proteins (DARPins, HEAT repeats, β‑propellers)	Modular, highly designable scaffolds; each repeat can be optimized independently, though interface compatibility must be preserved	Repeat unit engineering, interface redesign, loop grafting
β‑Barrel Outer‑Membrane Proteins	Pore‑forming proteins with defined strand‑strand hydrogen‑bond networks	Strand register engineering, pore size tuning, loop insertion
Fluorescent Proteins / FRET Sensors	Spectral tuning via chromophore environment; useful for biosensing	Chromophore pocket mutations, π‑stacking network, linker design
Self‑Assembling Protein Nanostructures (Cages, Filaments, 2D Arrays)	Engineered architectures for catalysis, drug delivery, nanotechnology	Symmetry‑driven docking, interface complementarity, rigid‑body design
Cytoskeletal Proteins (Actin, Tubulin)	Engineered polymerization dynamics, motor binding	Nucleotide‑binding loop mutations, subunit interface redesign, allosteric sites
Ubiquitin‑Proteasome System Components (E3 ligases, DUBs)	Targeted protein degradation, synthetic biology	E2‑E3 interface engineering, substrate‑recognition motif design, degron insertion
Artificial (De Novo) Enzymes	De novo scaffold with designed catalytic pocket	Rosetta enzyme design pipeline, transition‑state modeling, active‑site geometry optimization
Protein‑Protein Interaction Domains (PDZ, SH3, WW)	Small modular domains for peptide recognition	Loop grafting, interface redesign, combinatorial mutagenesis
Nuclear Receptors	Ligand‑dependent conformational change, DNA‑binding domain	Ligand‑binding pocket redesign, DBD‑effector domain fusion, allosteric modulation

Methodologies

Methodology	Mechanistic Core	Typical Use Cases
SELEX (Systematic Evolution of Ligands by Exponential Enrichment)	In‑vitro selection of nucleic‑acid aptamers based on binding affinity	Aptamer‑based sensors, targeted delivery
Phage / Yeast / Bacterial Surface Display	Genotype‑phenotype linkage via surface expression	Affinity maturation of antibodies, enzyme libraries
Error‑Prone PCR	Random mutagenesis across the entire gene	Diversifying libraries for directed evolution
DNA Shuffling / Family Shuffling	Recombination of homologous sequences	Creating chimeras with improved properties
Structure‑Guided Recombination	Crossover at structurally compatible positions	Domain swapping, scaffold optimization
Rosetta Design	Energy‑based optimization of side‑chain rotamers & backbone (including de-novo generation)	De novo protein design, interface redesign, enzyme active‑site engineering
AlphaFold‑Based Design	Predicting protein structures to guide mutagenesis or scaffold selection	Identifying tolerant regions for mutation, scaffold engineering
ProteinMPNN / RFdiffusion	ProteinMPNN for inverse folding & sequence generation from backbones; RFdiffusion for forward backbone generation	Generating novel backbones, conditional active‑site design
Molecular Dynamics (MD) with Enhanced Sampling	Mapping conformational landscapes, identifying hinge residues	Allosteric site discovery, gating mechanism analysis
Co‑evolutionary Analysis (DCA / EVcouplings)	Predicting residue‑residue contacts from multiple sequence alignments	Guiding mutagenesis to preserve network integrity
Genetic Code Expansion	Incorporation of non‑canonical amino acids via orthogonal tRNA/tRNA synthetase pairs	Introducing bioorthogonal handles, photocrosslinkers
Cell‑Free Protein Synthesis (CFPS)	In‑vitro translation of proteins without living cells	Rapid prototyping, testing toxic proteins, incorporating ncAAs
Phage‑Assisted Continuous Evolution (PACE)	Linking phage infectivity to a desired activity for continuous selection	Evolving enzymes, binding proteins, regulatory elements
Intein‑Mediated Splicing	Circularization or domain fusion via split intein splicing	Enhancing stability, creating cyclized proteins
PROTAC Design	Designing bifunctional molecules that bridge target to E3 ligase	Targeted protein degradation
Optogenetic Module Engineering	Mutating chromophore pocket or helix rotation to tune light response	Spatiotemporal control of activity
Allosteric Modulator Design	Creating pockets that bind small molecules to induce conformational change	Chemical control of enzyme or receptor activity
Scaffold‑Based Design (DARPins, Repeat Proteins)	Using repeat units as modular binding surfaces	High‑affinity binders, synthetic receptors
Synthetic Scaffolds (SpyTag/SpyCatcher)	Isopeptide bond between Asp side-chain and Lys ε-NH₂ within SpyCatcher for modular assembly	Enzyme cascades, multivalent display
Riboswitch Engineering	Designing ligand‑responsive RNA elements that control translation	Metabolic control, feedback regulation

Designability vs. Expressibility – Two‑Axis View

Protein Class	Designability (Structural/Computational)	Expressibility/Foldability (Cellular Context)
GPCRs	High – many structures, AlphaFold predictions	Medium – need membrane environment, chaperones
Ion Channels	High – pore architecture known	Medium – require lipid bilayers, proper gating
Enzymes	High – active‑site geometry often dictated by a few residues	Medium – some enzymes need cofactors or chaperones
Antibodies	High – phage/yeast display libraries	Medium – expression of full IgG can be challenging
Optogenetic Proteins	High – modular domains	Medium – chromophore incorporation
RTKs	Medium – dimerization and autophosphorylation complex	Medium – require extracellular ligand binding
Transcription Factors	Medium – DNA‑binding specificity can be altered	Medium – expression stability variable
Structural Proteins	Low – repetitive, β‑rich scaffolds	Low – hard to express recombinantly
Transporters	Medium – pore architecture known	Low – lipid‑dependent folding, assay difficulty
Viral Coat Proteins	Medium – capsid symmetry known	Low – assembly fidelity critical
Multi‑Domain Enzymes	Low – inter‑domain interactions tight	Medium – modular assembly possible
Highly Glycosylated Proteins	Low – glycoform control difficult	Low – folding depends on glycosylation
Humanized Antibodies	High – framework grafting is computationally tractable	Medium – expression of full IgG can be challenging
Scaffold‑Based Enzymes	High – repeat or domain scaffolds are designable	Medium – some scaffolds require chaperones
Azobenzene‑Controlled Proteins	Medium – incorporation site selection	Medium – photo‑isomerization may affect folding
Ubiquitin‑Tagged Proteins	Medium – degron design is tractable	Medium – need to avoid proteolysis

Other techniques to consider

Technique	Core Insight	Practical Implementation
Deep Mutational Scanning	Generates quantitative, position‑specific enrichment landscapes for folding, binding, catalysis, and cellular fitness (structural knowledge helps rationalise results)	Synthesize codon‑substituted libraries (including insertions/deletions), select under relevant conditions, NGS pre‑/post‑selection, compute enrichment scores.
Ribosome Display & mRNA Display	Enables selection from libraries limited only by synthetic DNA diversity, compatible with stringent or proteolytic conditions.	In‑vitro translation of ribosome‑arrested or puromycin‑tethered mRNA–protein fusions, affinity capture, RT‑PCR recovery, iterative rounds.
Thermodynamic Integration & Free‑Energy Perturbation	Alchemical MD protocols that mutate ligand atoms into dummy or alternate atoms, integrating ∂H/∂λ to yield relative ΔΔG (absolute binding requires double-decoupling or funnel-metadynamics)	Use enhanced‑sampling Hamiltonian replicas, accurate force fields, convergence diagnostics.
Kinematic Closure (KIC)	Solves protein loop closure analytically, enabling rapid enumeration of sterically allowed conformations that can be refined by rotamer repacking and energy minimization.	Apply to loop modeling in Rosetta or other modeling suites, integrate with MD for side‑chain optimization.
Native Chemical Ligation	Chemoselectively condenses an unprotected synthetic peptide bearing a C‑terminal thioester with another peptide containing an N‑terminal cysteine to form a native amide bond, enabling total chemical synthesis of proteins with precise labels (up to 300-400 aa via multiple ligations).	Synthesize peptide fragments, perform ligation under mildly acidic conditions, purify product, verify by MS.
Cystine‑Knot (Knottist) Engineering	Cross‑braces a short β‑sheet or helical core with three disulfides in a topological knot (cystine III-VI passes through II-V), imparting extreme proteolytic, thermal, and chemical stability while allowing hypervariable loops for recognition.	Design cystine‑knots in silico, express in systems that support disulfide formation, test stability.
Mirror‑Image Phage Display	Screen an L-peptide library against the synthetic D-enantiomer of the target (synthesised by solid-phase chemistry); the corresponding D-peptide of the selected binder is then synthesized to recognize the native L-target.
Algorithmic T‑Cell Epitope Mapping	Scans protein sequences for peptide epitopes whose length matches the anchor preferences for common MHC class II alleles, then redesigns surface residues to eliminate motifs while preserving structure and activity.	Use epitope prediction tools, apply conservative mutations, validate by immunogenicity assays.
Trinucleotide Mutagenesis	Uses pre‑synthesized phosphoramidite trimers corresponding to desired codons to build combinatorial libraries that exclude stop codons and allow user‑defined amino‑acid ratios at each diversified position.	Design oligonucleotide pool, perform PCR, transform, screen. Trinucleotide mutagenesis (e.g., Sloning method) explicitly uses pre-synthesized trinucleotide phosphoramidites during automated solid-phase oligonucleotide synthesis, which is a standard phosphoramidite-based chemistry process.
Hydrogen–Deuterium Exchange Mass Spectrometry (HDX‑MS)	Monitors time‑dependent deuterium incorporation into amide hydrogens as a readout of solvent accessibility and hydrogen‑bond stability; peptic digestion and high‑resolution LC‑MS/MS localize exchanged sites, enabling time-resolved conformational dynamics, allosteric pathways, and binding interfaces.	Perform HDX experiment, analyze MS data, map to structure.
Deep‑Learning Hallucination & Inpainting	Seeds a generative protein network with random noise or partial backbone coordinates, then iteratively optimizes a latent space under structural and functional constraints so that the network hallucinates complete protein backbones that cradle desired catalytic or binding motifs without relying on known folds (although it can still converge to known folds).	Use models like ProteinMPNN, Diffusion‑based generators, enforce constraints, validate by synthesis.
Fragment‑Based Drug Discovery (FBDD) Applied to Protein Active‑Site Inhibition	Identifies low‑molecular‑weight chemical fragments that bind subsites of the catalytic pocket with millimolar‑to‑micromolar affinity; structure‑guided linking or growing merges these fragments into larger ligands whose enthalpically favorable, vector‑directed interactions achieve nanomolar potency while maintaining ligand efficiency and selectivity.	Screen fragment library, co‑crystallize, design linkers, synthesize, assay.

excluded from the above table because it is not genetic:

Hydrocarbon stapling (chemical, not genetic) – Two α‑methyl,α‑alkenyl non‑canonical amino acids (e.g., (S)-pentenyl‑alanine and (R)-pentenyl‑alanine) are incorporated into a solid‑phase peptide, then the peptide is treated with a ruthenium‑based Grubbs catalyst under dilute, anaerobic conditions to perform a ring‑closing metathesis (RCM). The resulting all‑hydrocarbon bridge locks the two residues in a defined distance, dramatically stabilising an α‑helix and making the peptide resistant to proteolysis. Because the stapling step is chemical, it cannot be achieved directly in vivo; the stapled peptide must be synthesized, purified and then delivered (e.g., by cell‑penetrating delivery or microinjection).

Recommendations for Practitioners

Separate designability from expressibility – a protein may be easy to redesign but hard to produce in a given host.
Leverage structural data (X‑ray, cryo‑EM, AlphaFold) to identify tolerant regions for mutation and critical residues for active‑site redesign.
Combine rational design with directed evolution – use computational predictions to narrow library size, then apply phage/yeast display, PACE, or CFPS for high‑throughput screening.
Consider allosteric sites in addition to active sites; engineered allosteric modulators can provide tighter control and lower off‑target effects.
Employ modular scaffolds (repeat proteins, SpyTag/SpyCatcher, nanobodies) to create synthetic assemblies without destabilizing individual domains.
Use non‑canonical amino acids for site‑specific labeling, cross‑linking, or to introduce photo‑responsive elements (azobenzene).
Validate folding and function with biophysical assays (CD, NMR, SAXS, MD) before moving to cellular or organismal contexts.
Iterate: design → expression → characterization → feedback to model → redesign.

Common protein engineering design mistakes

Cysteine and Disulfide Management

Error: Deploying cysteine residues without defined pairing geometry or redox control, leading to misoxidation, disulfide scrambling, and covalent aggregation.

Correction: Position intentional disulfides only where structural context enforces correct pairing—typically with spacing appropriate for helical or β-hairpin geometry. For surface-exposed or unpaired cysteines, substitute with serine to preserve hydrogen-bonding capability, or with alanine/valine for buried positions where the thiol is not required. Note C-x₂-C is common in redox-active CXXC motifs where strain is functional.

Proline and Glycine Misplacement

Error: Inserting proline within α-helical or β-strand segments, destabilizing backbone hydrogen‑bond networks. Overusing glycine in positions requiring conformational rigidity introduces uncontrolled entropy.

Correction: Restrict proline to N-terminal caps, C-terminal caps, or tight turns where its fixed φ angle is compatible. Reserve glycine for flexible hinges, linkers, or turn regions; avoid consecutive glycine runs that create overly floppy segments.

Charged Residue Pitfalls

Error: Mismatching salt-bridge geometry (e.g., Asp–Lys pairs with suboptimal reach) or creating charge-clash magnets (sequential Arg-Arg-Lys or Asp-Glu-Asp; note poly-basic patches common in nucleic-acid-binding domains, stabilised by counter-ions) that can promote electrostatic repulsion and protease sensitivity.

Correction: Favor bidentate Glu–Arg or Asp–Arg pairs for robust salt bridges. Alternate charges (Arg-Glu-Arg-Glu) or insert neutral spacers (Gly/Ser/Thr) to avoid clustering. Account for local pKa shifts when using histidine as a pH sensor; for strong pH-independent charge, use Asp/Glu or Lys/Arg.

Post-Translational Modification Sequence Liability

Error: Retaining deamidation-prone motifs (Asn-Gly, Asn-Ser) that form cyclic succinimide intermediates, leading to backbone cleavage. Introducing unintended N-glycosylation sequons (Asn-X-Ser/Thr where X ≠ Pro) in loops, causing aberrant glycosylation.

Correction: Scan designs for NG/NS patterns and mutate Asn to Gln or Ala. If glycosylation is undesired, mutate sequons to Asn-X-Ala/Cys or rearrange loop length. Match the flanking sequence to the specific kinase consensus desired. Basic residues recruit PKA/PKC; Acidic residues recruit CK2; Proline recruits MAPK/CDK. To create sites for basophilic kinases (like PKA, PKC, CAMK) then use flank Ser/Thr/Tyr with basic residues and avoid embedding them in acidic blocks. If you flank with acidic residues, you target acidophilic kinases (like Casein Kinase II). If you flank with Proline, you target pro-directed kinases (like MAPK, CDK).

Oxidation and Chemical Degradation

Error: Exposing methionine or tryptophan to solvent, rendering them prone to irreversible oxidation.

Correction: Burial of Met/Trp is preferred; when surface exposure is unavoidable, substitute Met with leucine/isoleucine/norleucine, and Trp with phenylalanine to maintain hydrophobic packing while eliminating oxidation liability.

("norleucine" is a non-canonical amino acid rarely used outside specialized chemical synthesis, so its mention as a Met replacement is technically correct but practically niche.)

Hydrophobicity Mismatch

Error: Burying charged residues (Lys, Arg, Asp, Glu) in the core, destabilizing the fold. Conversely, leaving hydrophobic patches on the surface promotes aggregation.

Correction: Keep charged residues solvent-exposed (buried Lys/Arg tolerated if paired, e.g., in salt bridges). For core positions, use Leu, Ile, Val, Ala, Phe. Solubilize engineered interfaces by replacing exposed hydrophobics with Arg or Glu to enhance electrostatic solvation rather than using polar uncharged residues alone.

Metal Coordination Failures

Error: Recruiting histidine or cysteine for metal binding without proper spacing and geometry, resulting in weak or nonproductive chelation.

Correction: Adopt established coordination loops with appropriate residue spacing. Validate geometry against known structural templates; generic poly-histidine tags should include intervening linkers to prevent metal clustering.

Secondary-Structure Propensity Violations

Error: Placing β-branched residues (Val, Ile, Thr) in helical cores, causing steric clashes and backbone strain.

Correction: Favor leucine or alanine in helical cores. For β-strand edges, note that isoleucine and valine are common, whereas threonine is less favorable due to side-chain hydrogen-bond competition.

Linker Design Artifacts

Error: Using overly rigid linkers (Ala-Ala) that restrict domain movement, or over-relying on long flexible Gly4Ser repeats that risk O-glycan heterogeneity or proteolytic susceptibility in eukaryotic expression systems (risk mainly for serine-rich linkers with >3 Ser in a row).

Correction: Match linker type to function: flexible loops require Gly-Ser repeats with modest length; rigid domain separation benefits from Pro/Glu/Ala-rich sequences. Standard (G4S)n linkers are generally robust against xylosylation unless an SG context creates a cryptic Ser-Gly-Gly site, but long unstructured serine-rich linkers in eukaryotes (CHO, HEK293) risk O-glycan heterogeneity. Avoid runs of identical residues that create repetitive motifs.

Aromatic Residue Misuse

Error: Stacking aromatic residues (Phe-Phe-Trp-Trp) without interleaving charged/polar spacers, leading to π-stacking aggregation. Conversely, swapping aromatics without considering size, H-bonding, or redox potential disrupts packing.

Correction: Interleave aromatic clusters with charged or polar residues (Glu/Arg/Gln) and cap with soluble tails. Select Phe for pure hydrophobic packing, Tyr for H-bonding capability, and Trp for specific spectroscopic or stacking geometry requirements.

Aromatic stacking is sequence-dependent; Phe-Phe-Trp-Trp is common in antibody cores without aggregation.

Computational and Experimental Validation Gaps

Error: Assuming rational design rules are absolute without structural validation.

Correction: Employ computational modeling (Rosetta, FoldX) to preview clashes, calculate solvation energies, and validate motif geometry. Perform alanine scanning conservatively—use isosteric substitutions (Leu→Ile) rather than blanket Ala replacement to dissect contributions without obliterating structure. Optimize codon composition for the expression host to avoid rare-Arg codons that limit yield.

Protease and Immunogenic Sequence Liability

Error: Inadvertently introducing known protease cleavage motifs or immunogenic epitopes.

Correction: Scan final designs against protease specificity databases and epitope prediction tools. Avoid sequential recognition patterns and known antigenic combinations, especially in therapeutic candidates.

β-bulge Mis-labelling

Error: Listing Gly-x-Trp-x-Phe as a standard β-bulge.

Correction: Authentic bulges involve Gly/Pro/Asn at a single-residue insertion that shifts strand register, not a fixed W-F pattern. Apply established nomenclature: register shift by one residue with compensatory donor/acceptor geometry; design carefully to avoid strand register disruption.

Termini Traps

Error: Placing Pro, Asp, or Glu at extreme N- or C-termini; Met-Gly or Asp-Pro at ends prone to cyclisation or cleavage.

Correction: Keep Pro/Asp/Glu at least two residues inward; protect N-terminus with acetyl and C-terminus with amide if synthetic.

Met-Gly: This triggers Methionine Aminopeptidase (MAP) to cleave the initiator Met, leaving Gly at the N-terminus (cleavage event, not cyclization). Met-Gly is cleaved by MAP, and Map cleavage is a hydrolytic event, not cyclization. There is no cyclization risk for Met-Gly.

Asp-Pro: This peptide bond is acid-labile (cleaves during acidic elution or purification, but does not spontaneously cycle under physiological conditions like N-terminal Gln).

Cyclization: The true cyclization risk is N-terminal Gln (fast) or Glu (slow) forming Pyroglutamate.

The Asp-Pro peptide bond is acid-labile but does not spontaneously cyclize like N-terminal Gln/Glu.

His-tag Artefacts

Error: Relying on C-terminal His6 for solubility; tag may destabilise fold or bind metals (His-tag primarily for purification, not solubility).

Correction: Employ solubility partners (MBP, SUMO) or surface-charge engineering; position His-tag away from active site.

Redesigning an existing protein

Apply these rules to reprogram activity, specificity, stability, or regulation while preserving the ancestral fold of an existing protein: metal-ligand editing, gate-keeper resizing, dwell-time tuning, electrostatic Velcro, hydrophobic ratchets, modular proofreading, local rigidification, and incremental side-chain swaps.

First-Shell Ligands Control Chemistry, Not the Global Fold

A pair of Lewis‑basic side‑chains (Asp, Glu, His, Cys, or a backbone carbonyl) spaced ≈ 4–5 Å provides the electron‑pair donors needed to chelate one or two divalent cations (e.g., Zn²⁺, Mg²⁺, Ca²⁺). The metal ion itself is the Lewis acid, and the geometry of the basic ligands fixes it in place while leaving the surrounding scaffold essentially untouched.
Adding or deleting a single liganding atom (Asp→Asn removes a carboxylate; Ser→Cys introduces a thiol; Gly→Asp inserts a new carboxylate) typically lowers affinity but can change metal stoichiometry or geometry in engineered sites without perturbing the surrounding scaffold.
Because the rest of the domain is unchanged, the same fold can be tuned from fast-and-promiscuous to slow-and-stringent by altering only metal ligation.

A Single Steric Checkpoint Gates Substrate Scope

A bulky, often aromatic side chain that protrudes into the binding cavity acts as a size filter (many enzymes use multiple checkpoints such as oxyanion hole or second-shell H-bonds). Replacing it with alanine, glycine, or serine opens a cavity large enough to accept bulkier or chemically distinct substrates; restoring bulk reinstates stringency.
The gatekeeper does not need to contact the scissile bond; it only needs to block the entrance trajectory.

Selectivity Is a Kinetic Timer, Not an Equilibrium Constant

Stabilizing the “closed” conformation of a mobile loop or lid increases dwell time, giving more opportunities to reject incorrect substrates before chemistry occurs.
Inserting glycine or removing proline within hinges decreases the energy barrier between open and closed states, shortening dwell time and raising throughput at the cost of fidelity (Pro→Gly outcome system-dependent; can also increase dwell time by destabilising open state).
Pro→Gly or Gly→Pro mutations are therefore coarse switches for speed versus accuracy.

Electrostatic Velcro Tunes Residence Time Without Touching the Active Site

Clustering lysine or arginine on a surface ridge creates a positive patch that electrostatically traps a poly-anionic ligand, increasing processivity (effects screened in water to ~7 Å at 150 mM salt; additive within this range).
Neutralizing those residues (Lys→Gln, Arg→Ala) or adding same-charge repulsion converts a tight binder into a hit-and-release enzyme.
The effect is long-range (within ~10 Å) and additive; charge density rather than precise position matters.

Hydrophobic Ratchets Set Directional Preference

Leucine/isoleucine/valine ridges that interdigitate with a partner surface increase contact area (no strong evidence for strict directionality). Inserting glycine or small polar residues within these ridges introduces slip, allowing back-sliding or alternative trajectories.
Conversely, replacing glycine with branched aliphatics increases the energy cost of backward motion, biasing the reaction path.

Proofreading Modules Are Genetically and Kinetically Separable

A remote exo- or hydrolytic site can be silenced by a single carboxylate→amide mutation (Asp→Asn, Glu→Gln) that removes a metal ligand (may subtly perturb synthetic site via shared diffusible ligand such as nucleotide). The synthetic site largely remains intact because the two active sites communicate only through a shared diffusible ligand (usually a nucleotide or metal ion).
This uncoupling is general to any enzyme possessing an editing domain (synthetases, proteases, polymerases, transacylases).

Local Rigidification Beats Global Redesign for Thermostability

Introducing proline within surface loops or between secondary-structure elements (if φ/ψ compatible) reduces the conformational entropy of the unfolded state, raising the melting temperature without altering the active-site geometry.
Engineered salt bridges (Lys-Glu, Arg-Asp) at domain interfaces or at the N-cap of helices provide enthalpic stabilization with minimal structural risk.
Disulfide bonds can be installed between residues that are ≤6 Å apart in the native structure but distant in sequence; they rarely perturb folding if the χ1/χ2 angles are pre-compatible.

Sensing and Catalysis Can Be Genetically Uncoupled

Mutations in “trigger” or “switch” motifs that undergo order-to-disorder transitions can weaken substrate discrimination (specificity) without slowing chemistry, useful when intentional error incorporation is desired.
Conversely, tightening these motifs (e.g., replacing flexible linkers with short helices) creates ultra-high-specificity variants that still catalyze at wild-type rates.

Domain Modularity Permits Mix-and-Match Architecture

Because catalytic, lid, clamp, and editing domains are often connected by flexible tethers, they can be swapped, duplicated, or deleted to build new catalysts with altered processivity, substrate range, or error rates.
Peripheral insertions or deletions rarely perturb the core fold, providing a safe playground for targeting or stability modifications.

Side-Chain Chemistry Is a Tunable Continuum

Asp↔Glu, Asn↔Gln, Ser↔Thr, Lys↔Arg, Tyr↔Phe pairs offer steps in length, polarity, or H-bonding capacity (Asp→Glu can abolish H-bonding in constrained sites), allowing fine-gradient scans of function without structural shock.
Cysteine and histidine provide conditional control: cysteine can be oxidized to disulfide or labeled with maleimide probes; histidine can be protonated or metal-chelated to create pH- or redox-switchable gates.

Protein-Engineering Amino-Acid Motifs Cheat Sheet

Key Design Tools: Rosetta (energy-based design), FoldX (stability predictions), AlphaFold & ESMFold (structure prediction), ProteinMPNN (sequence design from backbones), PSIPRED (secondary structure prediction), MetalPDB (metal-binding sites). Validate motifs with PDB searches (e.g., RCSB PDB) and deep mutational scanning.

The rules below are heuristic guidelines requiring structural context, geometry validation (via MD/Rosetta), and experimental confirmation (e.g., CD, NMR). Propensities follow Chou-Fasman-style statistics but prioritize modern predictors like PSIPRED/AlphaFold.

Basic Amino-Acid Design Rules

Hydrophobic core: Leu, Ile, Val, Ala (FILVA), Phe, Trp, Met (aliphatic, often packs against aromatics), Tyr (amphipathic, common in cores with H-bonding hydroxyl).
Helix capping (Richardson & Richardson, 1988):
- N-cap: Asn, Ser, Thr, Asp – side-chain carbonyl oxygen acts as H-bond acceptor from exposed backbone N-H of residues n+2 or n+3 (e.g., PDB:1GCJ).
- C-cap: Gly (adopts left-handed alpha_L conformation to terminate helix and enable Schellman loop), Asn, Ser.
β-sheet edges: Polar residues (Asn, Ser, Thr) preferred at N-/C-terminal strand ends to reduce fraying; charged residues (Lys, Arg, Glu, Asp) used to prevent edge-to-edge aggregation via electrostatic repulsion; hydrophobic beta-branched residues (Ile, Val) better mid-strand.
Turns/Loops: Gly (flexibility), Pro (rigid kink), Asn/Asp/Ser (H-bonding).
Salt bridges: Asp/Glu ↔ Arg/Lys (i, i+4 or i, i+3 helical spacing preferred); buried bridges stronger but desolvation penalty must be compensated.
Disulfides: Cys-x(3–4)-Cys most common (e.g., immunoglobulin domains; spacing varies; oxidizing environment required (e.g., periplasm, eukaryotic ER).
Metal chelation (Martin, 1987; MetalPDB):
- Zn²⁺: C₂H₂ zinc finger with variable spacing, e.g., C-x(2-4)-C-x(12)-[H/x]-x(3-5)-H (tetrahedral; PDB:1ZAA); or Cys₄, Cys₃His.
- Fe²⁺/Fe³⁺: 2-His-1-carboxylate (2-His-1-Asp/Glu; octahedral; PDB:1AOR).
- Cu²⁺: His-x(3)-His (square planar; PDB:1RCY) or Met-rich axial sites.
- Geometry: ligands appropriately spaced; validate valence with MetalPDB.
Cyclization: β-hairpin stabilized by type I/II′ turns (e.g., D-Pro-Gly; PDB:1GB1); Trp-cage is an optimized 20-residue mini-protein fold in which a central Trp is caged by a single 3₁₀ turn, proline packing and a salt bridge, rather than a transplantable motif.
Dimer interfaces:
- Leucine zipper: Heptad repeats a-b-c-d-e-f-g with hydrophobic Leu/a at d-positions (parallel coiled-coil; e/g interchain salt bridges for specificity; PDB:1YSA).
- Knob-into-hole: Small side-chain (e.g., Val, Ala) knob fits into hole created by two large residues (reverse of large aromatic into pocket; e.g., PDB:1FBB).

Signature motifs and typical outcomes

Motif (Pattern)	Structural / Functional Pay-off	Typical Context / Example
G-x(4)-G-K-[T/S] (P-loop/Walker A)	Phosphate-binding loop for ATP/GTP (Lys contacts β/γ-phosphates; T/S hydroxyl coordinates Mg²⁺)	Kinases, GTPases
Asn-X-Ser/Thr (X≠Pro,Asp)	N-glycosylation sequon – targets oligosaccharyltransferase	Secreted/membrane proteins
Cys-x-x-Cys (CXXC)	Thioredoxin-fold redox motif (requires pKa modulation)	Thioredoxin, PDI
C-x(2-4)-C-x(12)-H-x(3-5)-H (C₂H₂)	Zn²⁺ tetrahedral coordination (classical zinc finger; spacing varies, e.g., C-x₄-C-x₁₂-H-x₃-H in Zif268)	Transcription factors
His-x(3)-His	Cu²⁺ / Ni²⁺ binding, often square-planar	Cupredoxins, His-tags
a-b-c-d-e-f-g heptad: hydrophobic a & d	Leucine zipper – parallel coiled-coil dimerisation (e/g charges give specificity)	bZIP TFs
Pro-x-x-Pro (PxxP)	Poly-Pro II helix for SH3 recognition (orientation set by flanking basic residue)	Adaptor proteins
Trp-x(7)-Trp-x-Trp	Aromatic cages (Trp, Phe, Tyr clusters) recognize methyl-Lys/Arg; spacing varies (check PDB for context)	Chromodomains, PHD fingers
Gly-x-Asn-Gly	β-turn (Asn-Gly type I'); stabilises β-hairpins (preceding Gly not canonical part of turn)	β-hairpin loops
D/E-x-D/E-x-D/E-G-[x]-x-D/E/N-x-x-E/D (EF-hand)	Ca²⁺-binding 12-residue loop (Gly-6 hinge; ligands at 1/3/5/7/9/12)	Calmodulin, C2 domains
Trp-x(2)-Phe-x(3)-Phe	Aromatic stacking pocket for hydrophobic ligands	β-barrel hydrolases
G-x-x-x-G (GxxxG) with G at i, G at i+4	TM helix dimerisation (small faces enable Cα-H···O bonds)	Glycophorin A
Gly-X-Y repeats (X/Y often Pro/Hyp)	Triple-helix stability in collagen (Gly every 3rd residue essential for tight packing; Pro/Hyp in X/Y promote PPII conformation)	Collagen

All motifs require precise backbone geometry; sequence alone insufficient (use ProteinMPNN for inverse design).

Engineering Hot-Spots

Target	Recommended Moves	Why
β-hairpin capping	Install Asn-Gly (type I') at turn (optionally with preceding Gly)	Asn side-chain staples strands via H-bonds; Gly enables φ/ψ flip (PDB:1GB1)
Helix N-cap	Place Asn/Ser/Asp at N1	Side-chain accepts H-bond from exposed backbone N-H of n+2/n+3, stabilises helix start (validate Rosetta ΔΔG)
Disulfide stitching	Introduce Cys-x(3–4)-Cys in loop/adjacent strands (Cβ-Cβ geometry compatible; Cys-x(2)-Cys strained but functional in redox motifs)	Clips unfolded entropy
Metal-site swap	Exchange ligands preserving field (e.g., Cys→His if tetrahedral; check MetalPDB)	Maintains valence/geometry; e.g., C₂H₂ → Cys₃His
Dimer interface	Enlarge buried chains (Ile→Leu→Phe) or add knob (small side-chain into hole)	Shape complementarity + buried SASA (Rosetta interface score)
Electrostatic Velcro	Cluster Lys/Arg on one face, Glu/Asp opposite (i, i+4 spacing)	Long-range zipping; cation-π with aromatics
Loop rigidification	Gly→Pro (or N+1 Pro) if Ramachandran allows	Fixes φ (~-60°), cuts unfolded entropy (AlphaFold confidence check)
Alanine scan	Bulky/polar → Ala (or isosteric: Leu→Ile, Asp→Asn)	Identifies linchpins; pair with FoldX ΔΔG
Oxidation shield	Met→Leu/Ile or Met→Nle, Trp→Phe, unpaired Cys→Ser/Ala (surface)	Preserves volume/H-bonds, removes ROS liability
Glycosylation veto	Asn-X-Ser/Thr → Gln-X-Ser/Thr or Asn-X-Ala (X≠Pro,Asp)	Blocks OST recognition (scan with NetNGlyc)

One-Letter Code Mnemonics

FILVA – Pure hydrophobics (Phe,Ile,Leu,Val,Ala); extend to FILMVWYAC for cores with Tyr amphipathic, Met oxidation-sensitive, Cys conditional on disulfide formation.
STNQ – Polar uncharged (Ser,Thr,Asn,Gln) – surface H-bonds.
DE – Negative (Asp,Glu) – surface or Ca²⁺/Zn²⁺ coord.
KRH – Positive (Lys,Arg,His) – nucleic acid binding, pH sensors (His).
GP – Loop/turn toolkit (Gly=flex, Pro=rigid).

Quick Designer's Checklist

Core: Minimize buried DE/KRH unless paired (e.g., salt bridges); hydrophobics inside (ProteinMPNN packing score).
Termini: N-cap Asn/Ser/Asp; C-cap Gly/Asn/Ser; avoid terminal Pro/Asp/Glu.
Redox: Remove unpaired Cys→Ser/Ala (cytosol); check with PDB disulfides.
PTM liability: Scan Asn-X-Ser/Thr (NetNGlyc), Asn-Gly/NS (deamidation).
Protease: Avoid motifs (e.g., furin RXXR; PROSPER).
Metal: Match coord./spacing (MetalPDB); Rosetta relax.
Preview: Rosetta/FoldX/ProteinMPNN for ΔΔG/clashes/solvation.
Host: Codon-optimize (e.g., IDT tool); avoid rare AGA/AGG (E. coli).
Immuno: NetMHC/IEDB for epitopes (therapeutics).
Validate: Tm (DSC), oligomer (SEC-MALS), activity; deep scanning for landscapes.