Abstract:
In forensic casework, DNA evidence profiles are generated by amplifying short tandemly
repeated lengths of DNA (STRs) using a polymerase chain reaction (PCR). DNA profiles can
be challenging to interpret when they are low-level, and there is the presence of more than one
individual in the DNA profile. DNA mixture interpretation can be subjective. However,
forensic scientists have developed methodologies to facilitate a more objective and consistent
approach to complex DNA profile interpretation over the past three decades. These are known
as probabilistic genotyping methods and many of them have been coded into software.
Recently, laboratories have considered adopting next-generation sequencing technologies
(NGS), also known as massively parallel sequencing (MPS), for forensic DNA profiling.
Sequenced-based information can increase the discriminatory power of STRs and have been
advertised as improving the resolution of profiles containing more than individual (called
mixtures). Few sophisticated probabilistic genotyping solutions are currently available for the
interpretation of NGS DNA profiles.
We explore, in this thesis, the current state of probabilistic genotyping for the interpretation of
forensic STR DNA profiles generated using capillary electrophoresis (CE) instruments
(Chapter 2). In Chapter 3, we describe how NGS DNA profiles are generated. Using our
understanding of continuous models for CE DNA profiles and how NGS DNA profiles are
generated, we develop models to calculate expected allelic and stutter signals in NGS DNA
profiles. The models developed are implemented within a probabilistic genotyping software,
and we test the performance of these models on the interpretation of mixtures in Chapter 4.
This body of work demonstrates that we can interpret DNA mixtures generated using a
sequencing technique. However, even though this solution is available, laboratories must
consider several other factors before implementing the technology in their workflow and
reporting the results.