Sequence alignment
Aligns two or more protein sequences and reports percent identity between them. Use to judge whether one protein is a usable template for another, to find conserved positions, or to check how similar two family members really are.
Sequence alignment
What it does
Takes several sequences saved in the project and lines them up so that corresponding positions sit in the same column. Reports the alignment itself and the percent identity of each sequence to the first.
Sequences come from the project's evidence — the UniProt search saves one FASTA per protein, and those are what you pick here. Two are required; up to four can be aligned at once.
How the alignment is made
Two engines are available and both answer the same question:
- Clustal Omega (the default) — builds a guide tree and aligns progressively. Reliable, and the usual choice for a handful of related proteins.
- MAFFT — faster on larger sets and often better where sequences differ in length or have large insertions.
They agree on straightforward cases and differ at the margins. If a result looks odd, running the other one is a reasonable check.
Percent identity is calculated over the columns where both sequences have a residue. Positions where either has a gap are excluded rather than counted as mismatches — otherwise adding one short sequence would drag every number down for a reason unrelated to similarity.
Reading the output
- Above 70% — closely related. One is usually a reasonable structural template for another.
- 30% to 70% — homologous but divergent. Conserved positions mean something; the rest need care.
- Below 30% — the "twilight zone". An alignment here is not reliable evidence of shared structure on its own, even when it looks plausible.
The aligned FASTA is saved so it can be opened elsewhere, alongside a summary of lengths, gaps and identities.
Important limitations
Sequence identity is not structural similarity. Proteins below 30% identity can share a fold; proteins above it occasionally do not behave alike.
This aligns sequences, not structures. If you have structures for both, a structural comparison is the stronger evidence.
Every sequence given is aligned, including one that does not belong. A distant outlier will distort a progressive alignment rather than being set aside.