An Integrative Statistical Framework for Assessing and Minimizing Errors in Ancient, Non-invasive and Forensic Genetic Studies
Molecular advances have made it possible to study wildlife populations by extracting and analyzing DNA from shed material such as hair or scat and from museum specimens. The familiar forensic practice of DNA fingerprinting can be used on samples of hair or scat to 'capture,' count and track individuals without ever handling or observing them. Unfortunately, these materials contain low quantities of DNA that introduce sporadic errors and inaccuracies into genetic data. Errors also occur when multiple individuals share the same DNA fingerprint or when one clump of hair, in fact, contains hair from two individuals. In this project, a set of statistical methods will be developed to assess the probability that a given sample contains each of the different possible errors. This leads to an efficient strategy for finding and removing errors by focusing further data acquisition on the most unreliable samples. Finally, the statistical framework will be packaged into a computer program and be made available to all researchers via the internet.
Forensic, non-invasive and historical genetic information on wild populations can be used to address questions that are impractical or impossible to address by non-genetic methods. This research project will make a major contribution to accessing this type of information because it directly addresses two issues of paramount importance, accuracy and efficiency. But just as with forensic DNA evidence in a court of law, this information must be reliable for the conclusions to be accurate. Because collecting genetic information is also expensive, its acquisition must be efficient.