TrueAllele solves 1963 Winnebago cold case using “inconclusive” DNA

Back to Presentations

Human factors in automated forensic DNA examination

M. W. Perlin, "Human factors in automated forensic DNA examination", Expert Working Group on Human Factors in Forensic DNA Interpretation: Probabilistic Genotype Software, National Institute of Standards and Technology, National Institute of Justice, 26-Jan-2021.


Talk

PowerPoint presentation of Dr. Perlin's talk.


PDFDownload Handout
PPTDownload PowerPoint

Abstract

The National Institute of Standards and Technology (NIST) and the National Institute of Justice (NIJ) formed an "Expert Working Group on Human Factors in Forensic DNA Interpretation" to better understand the role of human factors in forensic examination.

The NIST/NIJ Working Group invited Cybergenetics Chief Scientist Dr. Mark Perlin to speak with them about probabilistic genotyping software. Cybergenetics developed the TrueAllele® computer technology twenty years ago. The system identifies people in DNA mixture evidence by producing accurate likelihood ratio (LR) match statistics.

Dr. Perlin gave a 45 minute talk in January to the Working Group, followed by 45 minutes of Q&A. His talk began with some key principles of probabilistic genotyping:


  • PCR introduces randomness in amplified allele amounts
  • Genotype combinations explain DNA mixture data
  • Solving for all possible genotypes, up to probability
  • Bayesian modelling finds genotypes in STR evidence data
  • A two-step framework for objectively examining DNA mixtures
  • Genotype match distributions, validation studies, and error rates

The Working Group had asked Dr. Perlin eight specific questions. His talk continued by answering their questions. To paraphrase, the topics were:


  • How DNA complexity affects understanding and communication
  • Explaining probabilistic genotyping results to a jury
  • Validating a probabilistic genotyping system
  • Selecting DNA samples in a validation study
  • When do computers outperform a human analyst
  • How the LR depends on reference database assumptions
  • Any 'manufacturer's warnings' for TrueAllele crime labs
  • Automatic genotype comparison with contamination databases

These probabilistic genotyping principles and topics are of interest to forensic scientists, attorneys, judges, scholars, policy makers, statisticians, other scientists, and the general public. Dr. Perlin's 45-minute presentation to the NIST/NIJ Working Group is available on YouTube.