What is the difference between sequence identity and sequence similarity?

Why

Compute identity as identical aligned positions divided by a declared denominator, such as aligned columns or the length of a reference region. For proteins, similarity usually incorporates chemically tolerated substitutions from a matrix like BLOSUM, but software definitions of positives and percentages can vary. A leucine-to-isoleucine replacement can be similar under a matrix without being identical.

Always report alignment coverage, gap treatment and scoring method alongside percentages. A 95% identical 20-residue stretch is not interchangeable with 45% identity across an entire enzyme. For DNA, a simple exact-match identity is often more meaningful than an unspecified similarity percentage.

Neither metric independently establishes orthology, functional equivalence or evolutionary direction.