A new rule for analyzing homologous coding sequences in DNA.

A new rule is proposed for detecting homology between DNA sequences coding for proteins. Simple coding considerations predict that if two DNA sequences are homologous because of a common ancestry, they should share sequence similarities primarily in the same translation phase, with their codons aligned. Similarities which are in phase are shown to be more frequent between related sequences than between unrelated or random ones. But similar segments which are out of translation phase are no more frequent between related than unrelated sequences. Duplication and concatenation of genetic elements, followed by gene duplication and random mutations would lead to the patterns observed. Similarities are examined between various immunologically important sequences, including immunoglobulins, MHC products, and the T lymphocyte antigen Thy 1.