LikeLTD software

likeLTD (likelihoods for low-template DNA profiles)

likeLTD is an R package for computing likelihoods for DNA profile evidence, including complex mixtures and when profiles are subject to dropout.

See:

This gives information on installing and running, an illustrative analysis, a description of the model, validation study, version history and acknowledgments. Version 6.3 makes further improvements to the allele and output reports to simplify presentation of the crime scene profile (CSP) and known genotypes into a single table. The evaluation report now flags any parameter estimate that is close to the pre-specified bounds on the search space (the bound may need to be relaxed to achieve a better result). The default maximum heterozygous DNA contribution of an individual has been changed from 5000 to the maximum observed peak size in the CSP. A parameter to specify the minimum dropin value has been added. We include two new database files, covering 4 US populations for the Globalfiler and Identifiler kits. There is a minor bug fix in reporting rare alleles in the reports.

The above Guide only covers the new peak height model, for the discrete model (which is included in subsequent versions, essentially unchanged from Version 5.5), see the previous guide:

For an example witness statement illustrating how likeLTD results can be presented in court, see:

likeLTD is a team effort, early versions were coded by David Balding, Adrian Timpson and the UCL Research Software Development Team, in particular Mayeul d’Avezac, contributed much to versions 4 and 5, while the peak height model in Ver 6 has been developed and coded by Christopher Steele.

A publication describing comparisons of likeLTD (Ver 6.3) with another open-source software EuroForMix, finding that despite different modelling assumptions the two programs generate similar results, is available here:

A publication describing likeLTD Ver 6.0 is available here:

For Versions 5 through 5.5 the most relevant publication is:

which gives a description of the statistical model, and examples of analyses of real cases and results from simulations.  Other relevant publications:

Please send comments and report any bugs to Chris Steele at c.steele.11@ucl.ac.uk