When Peter Forster started compiling the world's most comprehensive database matching mitochondrial DNA (mtDNA) to geography, people laughed when he said he could tell from their saliva what their ancestors were doing 10,000 years ago.
This week, the Cambridge geneticist has suggested that, based on his mathematical modelling of mtDNA sequences, most published DNA sequences could be inaccurate. MtDNA is separate from cellular DNA and is passed down unchanged from mother to child. Unless a natural mutation occurs, it can be used to trace family histories over thousands of years.
In an editorial in Annals of Human Genetics , Dr Forster claims that more than half the mtDNA sequencing studies ever published contain simple copy errors. This undermines accepted research such as homogeneity of mtDNA of Icelanders or an African origin for mankind, and it has ramifications for all DNA sequences.
Dr Forster was born in London in 1967 and educated in Germany. As a chemistry undergraduate at Hamburg University, he was fascinated by Mitochondrial Eve, the hypothetical woman at the convergence of generations of human mothers. She was described in a paper by Dr Forster's hero, Berkeley researcher Alan Wilson, in 1987.
With a mathematician friend, Dr Forster worked on mtDNA sequences, trying to piece together how humans settled the continents and when. They stumbled across an algorithm that simplified the identification of plausible family trees from the data and began to build the database. In 1999, Dr Forster was invited to set up the molecular genetics laboratory at the Cambridge McDonald Institute for Archaeological Research. The database, featured in a BBC documentary last week, Motherland - A Genetic Journey , contains more than 20,000 mtDNA samples.
His ancestry-tracing service for the public will go as far back as Neanderthal times.