A statistical method for correlating tRNA sequence with amino acid specificity

A statistical method for finding the nucleotide positions in tRNA sequences that correlate with amino acid specificity has been developed. The procedure involves finding the subset of nucleotide positions and groups of positions where the marginal density of one amino acid tRNA class does not overlap that of any other amino acid class. The procedure is an application of a statistical method known as the Expectation Maximization algorithm.