Structural alignment

Content sourced from Wikipedia, licensed under CC BY-SA 3.0.

Structural alignment is a way to compare the three‑dimensional shapes of biomolecules, mainly proteins (and sometimes large RNAs), to see if they are related by structure rather than by sequence alone. It tries to match the overall fold without needing prior knowledge of which parts line up. This makes it useful for finding evolutionary relationships when the sequences are too different for standard sequence alignments.

What it gives you
- A 3D superposition: the structures are placed on top of each other as best as possible.
- A measure of similarity, usually a root‑mean‑square deviation (RMSD), which shows how far the structures differ after alignment.
- A set of equivalent residues (or atoms) that are considered matched between the structures.
- A translated one‑dimensional sequence alignment, from which you can still report sequence identity.

When and why to use it
- To compare two or more structures, especially when sequence similarity is low.
- To infer possible evolutionary relationships and classify protein folds.
- To evaluate predicted structures by comparing them with known, real structures.
- To help interpret data from structural genomics and proteomics, or to benchmark sequence‑only methods.

How it works in practice
- Because many proteins have multiple domains, care is needed: misaligned domains can inflate RMSD.
- The simplest and most common approach uses backbone atoms (usually alpha carbons) and ignores side chains unless the structures are very similar.
- Different methods use different ideas. DALI builds a distance matrix from fragments and finds the best match; CE uses aligned fragment pairs to build an optimal alignment; Mammoth focuses on statistically significant subalignments; TM‑align adjusts weighting to speed up and improve long‑range comparisons; SSAP uses vectors between neighboring residues.
- Some tools compare entire structures (global alignment), while others focus on local patterns (like binding sites).
- For large sets of proteins, all‑to‑all comparisons help classify folds or build databases. Local methods can also detect shared functional motifs.

RNA and special cases
- Structural alignment can be used for large RNA structures too, though the field is smaller.
- Some RNA methods decompose structures into smaller units to align and score them.

Choosing a tool
- There are many software packages with different strengths. Speed, the type of alignment (global vs. local), how scores are reported, and how well they handle flexible regions all matter.
- In practice, researchers often use more than one method and compare results to gain confidence.

Limitations
- No single method is best for all situations. Different scoring choices and representations can lead to different conclusions.
- Highly similar sequences can be aligned easily, but remote homologs require careful interpretation.
- The quality of the input structures (experimental or predicted) affects the results.

Bottom line
Structural alignment helps us see when two biomolecules share a similar shape, suggesting a common ancestry or a conserved function, even when their sequences look very different. It produces a coordinate overlap, a measure of similarity, and a map of corresponding parts, guiding insights into biology that sequence data alone might miss.