Using computer modeling to analyze DNA evidence

Lun-104For a research project that sounds like it could have been taken from a CSI script, Desmond Lun, an associate professor of computer science at Rutgers–Camden, is working with a team of scientists from Boston University and MIT to use signal-processing techniques to determine a suspect’s likelihood of contributing to a DNA sample.

Lun was recently awarded a new grant from the National Institute of Justice to continue the research he started last year.

“We use advanced signal-processing techniques to directly determine how likely a suspect is to have contributed to a sample,” Lun explains. “These techniques are particularly important in cases where DNA samples are mixed (i.e. where other people have contributed to the sample) and in cases where only a small amount of DNA can be obtained.  In these cases, existing techniques can easily give the wrong likelihood–implying a much greater or much lower probability of a suspect contributing than is in fact the true.  Our techniques will rectify this problem, giving a more accurate determination of this probability.”

DNA can be found in any human cells including blood, hair, and skin. Part of the project is to develop free software that crime labs can use to test the new method. If they find it useful, the software could gradually become incorporated into standard practice.

Lun teaches courses in computational and integrative biology at the Camden Campus of Rutgers, The State University of New Jersey. He received bachelor’s degrees in mathematics and computer engineering from the University of Melbourne, Australia and earned his master’s degree and doctorate in electrical engineering and computer science from MIT.

Posted in: Research Highlights

Comments are closed.