Non-random relationships among amino acids in protein sequences.

We have computed the observed distribution of amino acid pairs n, n+1 up to n, n+10 within 100 unrelated protein sequences containing a total of 14,034 amino acids. The expected distribution of the same amino acid pairs was calculated assuming that the amino acids within each sequence are arranged in random order. Statistical analysis of the data shows that amino acids in the n, n+1 and n, n+3 relationship deviate from random expectation at the 1% level of significance. These relationships are consistent with known secondary structures; however, there are no amino acid pairs which obviously contribute to the non-random result.