The exploitation of assembly language instructions in biological text manipulation: I. Nucleotide sequences

Abstract

We explore the numerical interpretation of biological texts with a view to exploiting the efficiency with which digital computers manipulate binary strings. The central feature of our proposition is that not all numerical interpretations of biological text are digitally equivalent. Certain specific biological to numerical text mappings permit the exploitation of assembly instructions for processing entire strings in a single instruction cycle, thereby avoiding expensive digit by digit manipulation for nucleotide or amino acid sequences. It is shown that the choice of mapping from biological to numerical text can critically effect efficiency. For nucleotide texts, we find that, of the 24 possible bijective mappings from the nucleotide alphabet to the quaternary number system, one subset of eight mappings corresponds to the interpretation of polymerase as a base-4 3's complement arithmetic operator, allowing polymerase to be particularly efficiently modelled using the assembly language NOT instruction.