An ab initio quantum chemical investigation of the error-coding model of nucleotide alphabet composition

Abstract

In the error-coding model of nucleotide alphabet composition [MacDónaill, D. A., 2002, Chem. Comm., 2062], the recognition features of nucleotides are mapped to 4-digit binary numbers, capturing the hydrogen acceptor/donor patterns (3 bits) and the purine/pyrimidine size motif (1 bit). Optimal alphabets, from the perspective of fidelity, are those corresponding to parity check codes in which the purine/pyrimidine feature relates to the acceptor/donor pattern as a parity bit. The validity of the model depends on a broad parallel between the interaction energies of potentially associating nucleotides and a measure of complementarity, namely a Hamming type distance derived from the XNOR product of the corresponding binary interpretations. In this paper we model both complementary and non-complementary purine—pyrimidine associations at the ab initio 6–31G* level of approximation. The results suggest that the 4 bit representation of nucleotides, on which the error-coding model is based, adequately captures nucleotide interactions for the purposes of considering the composition of the alphabet. The results also indicate that a single mismatch in H-bonding positions between non-complementary nucleotides is insufficient to ensure replication fidelity, which requires that non-complements mismatch in two of three positions, a requirement satisfied by a parity code structure.