Compare Similarities Between DNA Sequences Using Permutation-Invariant Quantum Kernel

Abstract: Computing the similarity between two DNA sequences is of vital importance in bioscience. However, traditional computational methods can be resource-intensive due to the enormous sequence length encountered in practice. Recently, applied quantum algorithms have been anticipated to provide potential advantages over classical approaches. In this paper, we propose a permutation-invariant variational quantum kernel model specifically designed for DNA comparison. To represent the four nucleotide bases in DNA sequences with quantum states, we introduce a novel, theoretically motivated encoding scheme: the four distinct bases are encoded using the states of symmetric, informationally complete, positive operator-valued measures (SIC-POVMs). This encoding ensures mutual equality: each pair of symbols is equidistant on the Bloch sphere. Also, common DNA similarity measures, such as the Levenshtein distance, exhibit permutation-insensitive property. To approximately capture this property, we specially design a parameterized quantum layer to realize permutation invariance in the kernel model. We show that our novel encoding method and parameterized layers used in the quantum kernel model can effectively capture the symmetric characteristics of the pairwise DNA sequence comparison task. We validate our model through numerical experiments, which yield promising results on length-$8$ DNA sequences.