Levenshtein distance as a measure of accuracy and precision in forensic PCR-MPS methods

Accuracy and precision determinations are standard components of method validations where they help to describe the performance of methods. Despite their importance, a standard approach to calculating these parameters is not available for the DNA sequence property of samples in forensic PCR-MPS methods. DNA sequence is a nominal property lacking magnitude and therefore standard approaches to expressing accuracy and precision do not apply. However, sequence determinations by PCR-MPS methods can be expressed in terms of Levenshtein distance, which is a numeric metric to which standard statistical calculations can be applied. We illustrate how Levenshtein distance expresses method accuracy in terms of the closeness of sequencing reads to template DNA sequences, and method precision in terms of the agreement among sequencing reads. By expressing method performance in terms of a numeric metric, this method places PCR-MPS on equal footing with numeric measures of accuracy and precision in PCR-CE methods. Summary statistics based on Levenshtein distance can be used to compare performance of different forensic kits, markers, or methods. Levenshtein distance for expression of accuracy and precision of the DNA sequence is contrasted with the alternative measures phred score, Bhattacharyya distance, Shannon's entropy, and Gibbs' index of qualitativevariation (M1).

Automated summary

This study discusses how accuracy and precision in DNA sequencing can be expressed using Levenshtein distance, and how this method can be compared with alternative measures.