Sequence first. Structural validation next.
Find structural
neighbors.
Search AFDB50 with a protein sequence. Add its structure to align the candidates with DALI.
01 / Your query
Start with a sequence
Search workspace
Your search
- 1Encode sequence
- 2Find candidates
- 3Match structure
- 4DALI alignment
Query & search provenance
02 / Embedding search
ANN candidates
Ranked by embedding cosine similarity. These candidates have not yet been validated by structural alignment.
| Rank | Model | Cosine | Length | DALI status | Structure |
|---|
Continue your search
Add the query structure
Supply coordinates for the same protein sequence. The selected chain must match your query before structural comparison begins.
03 / Structural comparison
DALI alignments
Z ≥ 2 is the DALI significance threshold. Missing reports and failed comparisons have no score. Expand a model to inspect alignment intervals.
| Rank | Model / alignment | Z-score | RMSDÅ | Alignedresidues | Querycoverage | Targetcoverage | Seq. identity | Status |
|---|
Keep the evidence
Results & provenance
Available outputs include candidate ranks, alignment records and the actual search identity.
The method
Two distinct stages.
One traceable search.
- 01
Sequence → candidates
DaliH encodes the full query sequence and retrieves neighbors from the admitted AFDB50 index. Cosine similarity measures embedding proximity.
- 02
Coordinates → alignments
DALI compares a matching query structure against selected candidates. Its Z-scores and residue alignments provide separate structural evidence.