Issue No 7: November 2003
Special fund for SARS research
Q&A: The purpose of monitoring
Work on forensic DNA improves clarity of the probability factor
A closer look at meromorphic functions
Optimisation for production schedules
Reducing interference on mobile phones
Mathematical theory of fluids gives designers data on virtual models
Equation that can predict spots on a seashell
Research shows that filters for sound and images are correct
Short cut to finding best delivery route

Statisticians at The University of Hong Kong have been helping to solve crime scene complications where potential evidence is stained with DNA from more than one person.
Previously, the probability of a match between mixed DNA stains found at a crime scene and samples taken from a suspect later found to be innocent has been described as “unlikely” or “very unlikely.”

Readout of a mixed DNA stain from a crime scene with samples from the victim and suspect for comparison (above right) Prof Fung with DNA model (above left)

Now law enforcers can describe probability as a statistical value thanks to a research project headed by Principal Investigator, Prof Tony Wing-kam Fung. For a positive identification, forensic investigations involving DNA often need at least nine matches at sites, or loci, identified at specific points across the 23 pairs of chromosomes found in humans.
Prof Fung drew data from DNA profiling systems used by police in Hong Kong, Beijing and Taiwan to compare the frequencies of allele, or distinct types of DNA, at each of these loci.
“We found the frequencies were almost the same in all three cities,” he said. All loci are “non-coding” which means they do not describe the colour of skin, hair, eyes, or ethnic orientation. Prof Fung went on to verify the extent that the loci were independent. He said: “The nine loci used for identification cannot be 100 per cent unrelated because human beings are not all mated randomly.”
“We wanted to find out which sites were the least related. The more unrelated the sites are, the easier the calculation.” With mixed DNA stains from a crime scene, samples found by forensic detectives may include DNA from both the victim and perpetrator.
“As in principle we are all unique, except for identical twins, the fragment length of the loci of various individuals should be different,” said Prof Fung.
“In mixed stains, we can identify the DNA fragments contributed by the victim which leaves the remaining DNA fragments for comparison with the suspect.” Prof Fung’s research has developed the statistical theories and calculations to help in the analysis of mixed stains.
The theorems he obtained can be used in cases where more than 10 people are involved, as well as for multi-ethnic groups and relatives of suspects where DNA similarities may exist.
Some of the principles are being used in paternity testing and is the basis of a computer application, EasyDNA, developed by Dr Fung. To date, the software has been used by laboratories in Asia, Europe, North America and Latin America.

Principal Investigator
Prof Tony Wing-kam Fung :