Single-sequence protein structure prediction

What is single-sequence protein structure prediction?

Single-sequence protein structure prediction is the computational process of building a three-dimensional protein model from one amino-acid sequence without requiring a supplied multiple sequence alignment or template. Modern methods obtain structural signal from protein language models, but accuracy remains sequence- and region-dependent.

These models trade explicit user-supplied evolutionary input for representations learned from large sequence collections. They reduce setup time and are useful when MSA construction is slow or homologs are sparse.

No supplied MSA does not make a model evidence-free: training data still informs the representation. Compare predictors, retain native confidence outputs, and inspect domain packing, termini, and flexible regions carefully.

When to use single-sequence protein structure prediction

Use a sequence-only predictor when an MSA is unavailable or unnecessary for the first pass:

• A protein has few detectable homologs.
• You need a fold hypothesis for construct or domain planning.
• Several model families can be compared on the same sequence.

Strengths and limitations

One reviewed FASTA can produce a structural hypothesis without a separate alignment-search workflow. Running ESMFold, ESMFold2 and MiniFold side by side shows whether a region or domain arrangement depends on the model family.

The predictors still inherit training-set knowledge, and local confidence does not guarantee correct domain packing. Flexible termini, long linkers and oligomer-dependent regions are common sources of ambiguity. A single-sequence model does not establish the biologically relevant state.

Applications of single sequence protein structure prediction

Single-sequence predictors are useful when an MSA search is unavailable, slow or uninformative for an orphan protein. They can supply a first fold hypothesis for construct design, domain annotation or further structural analysis.

Comparing several model families can reveal whether a domain arrangement is stable across assumptions. Similar coordinates alone do not validate a novel fold, especially for disordered segments or oligomer-dependent regions.

How to run single-sequence protein structure prediction online

1. Prepare one sequence. Confirm construct boundaries, ambiguous residues and whether the biological state is monomeric.
2. Run the predictor panel. Generate models with ESMFold from the same FASTA.
3. Compare newer and smaller models. Review ESMFold2 outputs and the separate MiniFold branch without merging their native scores.
4. Inspect local confidence. Mark uncertain loops, termini and interdomain orientation before choosing a model.
5. Export alternatives. Keep all structures, native confidence files and model settings for downstream validation.

How to interpret single sequence protein structure prediction results

Local confidence can identify regions a model considers uncertain, but it does not establish the relative orientation of domains or the biologically relevant conformational state. Compare structures after alignment and report disagreements explicitly.

A predicted fold supports further investigation; it does not measure stability, binding, activity or function.

How single-sequence protein structure prediction works

Submit one FASTA sequence to ESMFold, ESMFold2, and MiniFold and compare their structures and confidence.

1. Clean the sequence. Provide one ungapped amino-acid sequence with the correct biological boundaries.
2. Run language-model predictors. Run ESMFold, ESMFold2, and MiniFold from the shared workflow input.
3. Collect native confidence. Download every model and retain pLDDT, PAE, or other native confidence outputs when returned.
4. Compare model agreement. Align the predictions and inspect domain packing, termini, loops, and low-confidence segments.
5. Select regions for validation. Choose a model or ensemble for the downstream question and record the selection rationale.

Inputs and outputs

Check formats before running, then inspect and download the result from every workflow step.

Inputs

Research input

FASTAPDBmmCIF

One protein sequence in FASTA or plain-text format.

Outputs

Prediction and review outputs

PDBmmCIFCSVJSON

Predicted PDB or mmCIF structures with model-native confidence and downloadable files.