How to use the Interactive Construct Builder

Use the builder to inspect and adjust antigen boundaries from each protein report. Select a construct or enter start and end positions, inspect the available evidence, then copy its sequence.

What the builder shows

Reports with a compatible local AlphaFold model include the interactive builder and Live Selected Region table. The pLDDT trace and PAE matrix appear when their data are available. Reports without a compatible model retain precomputed construct cards and their sequence exports.

Selections stay synchronized across the panels present in the report. Selecting residues in the sequence or plots highlights the same residue window in the structure and updates the TSV/FASTA outputs.

Basic workflow

Start from the construct classes present in the report: full design region, PDB-backed constructs, strict calculated regions, lenient calculated regions, then annotated domains.
Choose a construct, enter start and end positions and click Apply boundaries, or select residues in the sequence or available plots.
Inspect the model and annotations at both boundaries.
Review warnings for unpaired cysteines and furin-like cleavage sites.
Copy the TSV or FASTA output for cloning, ordering, or downstream analysis.

Live Selected Region

The table reports sequences for the current selection: human construct name, boundaries, sequence, and equivalent mouse and cynomolgus monkey regions where ortholog mappings exist.

The construct name follows the format ENTRY_SPECIES_start-end, where ENTRY is the UniProt entry-name prefix (for example, PD1L1 for PD-L1, gene CD274). Selecting optional Cys-to-Ser mutations adds mutation suffixes to the name and updates the sequence immediately.

Structure panel

Reports with a local AlphaFold model show the structure as a cartoon colored by AlphaFold confidence. Selected residues are highlighted for boundary review against folded domains, linkers, cysteines, and visible surface regions.

Cysteines are emphasized in yellow with side chains where present. Exposed or unpaired cysteines can complicate recombinant soluble antigen expression and purification.

How to interpret pLDDT

pLDDT is AlphaFold's per-residue local confidence estimate. It describes confidence in the immediate geometry around a residue. Domain-domain orientation requires PAE review.

pLDDT range	Practical interpretation	Construct-design value
90-100	Very high local confidence.	Check these residues against domain and topology annotations when choosing boundaries.
70-90	Generally confident local structure.	Often acceptable for domain cores and boundary-adjacent residues.
50-70	Low confidence or flexible geometry.	Review manually. This range often marks flexible linkers, uncertain loops, or boundary regions.
Below 50	Very low confidence, often disordered.	Consider trimming unless the region has a required biological role.

For soluble antigen design, high mean pLDDT and a high fraction of residues above the structured threshold support local fold confidence. Use PAE to evaluate relative orientation between domains.

How to interpret PAE

PAE, predicted aligned error, estimates how uncertain AlphaFold is about the position of one residue or region when another residue or region is aligned. Lower PAE means AlphaFold expects the relative placement of those residues to be reliable; higher PAE means the relative placement is uncertain.

PAE pattern	What it suggests	Construct-design value
Low PAE block within a region	Low predicted error within the region.	Compare the block with annotated domains before choosing a construct.
High PAE between two neighboring blocks	Uncertain relative placement of the blocks.	Consider splitting constructs at or near the linker.
Low pLDDT and high boundary PAE	This is a candidate boundary between regions.	Review as a possible trimming or domain-separation point.
High pLDDT but high inter-domain PAE	Each domain has local fold support, with uncertain domain orientation.	Expression and relative domain orientation require experimental testing.

PAE can differ when the aligned and evaluated residues are reversed. It describes model uncertainty; it does not measure the stability of an expressed construct.

Choosing boundaries

Inspect both termini and avoid cutting through a helix, strand, or required domain. A flexible linker may provide a suitable boundary. For soluble constructs, review whether signal peptides, transmembrane segments, cytoplasmic tails, or disordered extensions should be excluded, and account for annotated processing sites.

Compare candidate boundaries with domain annotations, PDB coverage, and processing sites. Use pLDDT and PAE to inspect the model near each cut.

Strict versus lenient constructs

Strict constructs start from pLDDT segments and may be split further using PAE. Lenient constructs use a lower pLDDT threshold and allow longer gaps. See Constructs for the thresholds and filters.

Warnings

Cysteine warnings report missing or ambiguous partners in the model. PTM annotations identify recorded modifications and processing sites. Furin warnings mark sequence motifs for review; a motif match does not establish cleavage in your expression system.

Manual overrides

Use manual boundary choices for ligand-binding sites, obligate partners, required disulfides, epitope constraints, glycosylation, and literature-supported constructs.

Practical checklist

Confirm the intended topology: soluble extracellular region or membrane protein.
Inspect pLDDT across the selected sequence.
Check PAE within the region and across its boundaries.
Review terminal residues and any linker retained in the construct.
Check cysteine partners, furin motifs, and annotated binding sites.
Check the selected sequence against PTM and processing annotations.
Do mouse and cynomolgus monkey equivalent sequences preserve the selected region if cross-species antibody screening is planned?

For the full methodology behind the pre-generated construct list, see the Pre-generated Constructs page.