StructomeDB

v1.0 · 2026

Vanessa RobinsonHaoyue HuAshar J. Malik*  and  David B. Ascher*

Structome-wide pairwise protein structural and sequence comparison database

104,528,817 structurally detected pairs from an exhaustive all-vs-all comparison of 61,631 UniProt-linked representative chains from the full experimental PDB, spanning all domains of life including viruses. Each pair carries a unified feature vector of structural and sequence alignment metrics with full biological annotation.

104.5M unique pair records
61,631 representative chains
7,351 species represented
All domains covered including viruses
Sequence–Structure Similarity Landscape
Sequence similarity (BLASTP)
Structural similarity (Foldseek TM-score)
Threshold: struct ≥ 0.5  ·  seq ≥ 0.50  (click or drag to move)
++ Homologs — high structural & sequence similarity
+− Convergent candidates — high struct, low seq
−+ Divergent candidates — low struct, high seq
−− Unrelated — low structural & sequence similarity
Quadrant Counts
++ Homologs
High struct · High seq
+− Convergent
High struct · Low seq
−+ Divergent
Low struct · High seq
−− Unrelated
Low struct · Low seq
Export

10,000 pairs sampled (if available) per quadrant as defined by the user.

Rate limited to 5 exports / minute · 20 / hour per IP.
Compare two proteins

On-demand structural and sequence comparison

Analyse →
Two-axis classification
Each pair is classified by Foldseek TM-score (structural) and BLASTP similarity (sequence) into four quadrants across a 10×10 grid with 0.1 bin resolution.
Full feature vectors
Every pair stores alignment coordinates, rotation matrix U and translation vector T, enabling instantaneous structural superimposition without re-running any search.
Analyse any pair
Submit any two PDB chains via the Analyse endpoint to compute the full pairwise feature vector on demand, regardless of whether the pair is in the pre-computed database.