Much DNA evidence goes unused. Sophisticated crime laboratories can produce superb data that human analysts cannot interpret. DNA that is mixed (two or more contributors), damaged (degraded into pieces), or low level (with few molecules present) introduces uncertainty that overwhelms human problem solving. Thus, vital evidence in crimes of rape, murder, or terror is often understated or pronounced "inconclusive." Public safety is compromised when DNA identifications are not made, criminals remain at large, and preventable crimes are committed.
Mathematical modeling can embrace this uncertainty, expressing data variation in probability equations. Solving these high-dimensional problems with Markov chain Monte Carlo (MCMC) statistical search, a MATLAB® program can explore many thousands of scenarios. The process mathematically separates mixture data into the genetic components, or genotypes, of the contributors. Comparing genotype probability distributions calculates a DNA match statistic that can scientifically associate crime scene evidence with criminals.
This session describes the scientific development and forensic application of Cybergenetics TrueAllele® Casework, a computer system for interpreting and matching complex DNA evidence. Validation studies show that TrueAllele is more accurate than human data interpretation, preserving information from discarded evidence. TrueAllele supercomputer automation reanalyzed the World Trade Center DNA data to identify victim remains. A hundred TrueAllele reports have been filed on DNA evidence in serious crimes, while high-tech TrueAllele databases make cold hits to find investigative leads. TrueAllele mines existing crime lab data and completes the DNA process by computing a safer and more just society.
