Sequence Identity Calculator

Calculate percent identity, similarity, and evolutionary distance between biological sequences.

Length of aligned sequences (bp or aa)

Number of identical positions

Number of similar positions (conservative substitutions)

Number of gaps

Please check your inputs and try again.

Identity, Similarity and Why They Differ

Two sequences can be compared at different levels of strictness, and the terms are not interchangeable.

Measure	Counts	Typical relationship
Identity	Exact matches only	Always the lower figure
Similarity	Matches plus conservative substitutions	Higher — leucine for isoleucine counts
Coverage	Fraction of length aligned	Independent of both

Similarity uses a substitution matrix such as BLOSUM62, which scores biochemically conservative swaps positively. A leucine-to-isoleucine change preserves hydrophobicity and usually structure, so it counts toward similarity but not identity.

The Twilight Zone

Identity	Interpretation	Confidence
> 70%	Almost certainly same function	Very high
30–70%	Homologous, likely related function	High
20–35%	Twilight zone	Ambiguous
< 20%	Cannot infer homology from identity alone	Low

Below roughly 30% identity, sequences that share structure and function become statistically indistinguishable from unrelated pairs. This is why E-value matters more than percent identity for distant relationships — it accounts for alignment length and database size, so a 25% identity match over 400 residues can be far more significant than 45% over 30.

Coverage is the other essential number. High identity over a short stretch may reflect a shared domain rather than overall relatedness, so identity should always be reported alongside the aligned length.

Worked Examples

Example 1: Closely related proteins: 270/300 identical

Identity=90%, similarity=95%

Result: Highly conserved — likely same function across species

Orthologs in closely related organisms

Example 2: Distantly related: 90/300 identical

Identity=30% — twilight zone

Result: Structural homology may still exist

Below 30%: impossible to assign homology by sequence alone

Example 3: Why E-value beats percent identity

25% identity over 400 residues versus 45% over 30 residues

Result: The longer alignment is far more significant

A short high-identity match occurs by chance readily. E-value accounts for alignment length and database size, which raw percentage does not.

Example 4: Identity versus similarity

Alignment with 40 identical and 25 conservative substitutions in 100 positions

Result: 40% identity, 65% similarity

Conservative substitutions preserve biochemical character. The gap between the two figures indicates how much divergence is structurally tolerated.

Common Mistakes

⚠️

Reporting identity without coverage

High identity over a short alignment may reflect one shared domain. The aligned length is essential context for interpreting the percentage.

⚠️

Trusting percent identity in the twilight zone

Below about 30%, homologous and unrelated pairs overlap statistically. E-value or structural comparison is needed to distinguish them.

⚠️

Confusing identity with similarity

Similarity counts conservative substitutions and is always higher. Quoting one while meaning the other overstates or understates relatedness.

⚠️

Ignoring the substitution matrix used

BLOSUM62 suits moderately diverged sequences; BLOSUM45 is better for distant relationships and BLOSUM80 for close ones. The choice changes the alignment.

Frequently Asked Questions

Identity vs similarity?

Identity: exact same amino acid/nucleotide. Similarity: conservative substitution (similar chemical properties, e.g., Leu→Val both hydrophobic). Both count toward evolutionary distance, but identity is more stringent.

BLAST E-value connection?

E-value = expected alignments by chance in database of that size. E<0.001: significant. E<10⁻²⁰: highly significant. Related to but not identical to percent identity — alignment length and database size matter.

What is the difference between identity and similarity?

Identity counts only exact matches. Similarity also credits conservative substitutions such as leucine for isoleucine, so it is always the higher figure.

What is the twilight zone?

Roughly 20–35% identity, where homologous and unrelated sequences become statistically indistinguishable by identity alone. Structural or profile methods are needed.

Why does E-value matter more than percent identity?

Because it accounts for alignment length and database size. A modest identity over a long alignment is far more significant than high identity over a short one.

Which substitution matrix should I use?

BLOSUM62 is the general default. BLOSUM45 performs better for distantly related sequences and BLOSUM80 for closely related ones.