Effect of multiple allelic drop-outs in forensic RMNE calculations

Abstract

Technological advances such as massively parallel sequencing enable increasing amounts of genetic information to be obtained from increasingly challenging samples. Certainly on low template, degraded and multi-contributor samples, drop-outs will increase in number for many profiles simply by analyzing more loci, making it difficult to probabilistically assess how many drop-outs have occurred and at which loci they might have occurred. Previously we developed a Random Man Not Excluded (RMNE) method that can take into account allelic drop-out while avoiding detailed estimations of the probability that drop-outs have occurred, nor making assumptions about at which loci these drop-outs might have occurred. The number of alleles that have dropped out, does not need to be exactly known. Here we report a generic Python algorithm to calculate the RMNE probabilities for any given number of loci. The number of allowed drop-outs can be set between 0 and twice the number of analyzed loci. The source code has been made available on https://github.com/fvnieuwe/rmne. An online web-based RMNE calculation tool has been made available on http://forensic.ugent.be/rmne. The tool can calculate these RMNE probabilities from a custom list of probabilities of the observed and non-observed alleles from any given number of loci. Using this tool, we explored the effect of allowing allelic drop-outs on the evidential value of random forensic profiles with a varying number of loci. Our results give insight into how the number of allowed drop-outs affects the evidential value of a profile and how drop-out can be managed in the RMNE approach.