Genome data called into question

Craig Venter heads biotech firm Celera Genomics

By BBC News Online's Ivan Noble

A new study says that there may have been serious discrepancies in the way the US company Celera Genomics sequenced and interpreted the genome of the fruitfly, Drosophila melanogaster.

I think the publication was a bit rushed.

Samuel Karlin
Stanford University

The charge has implications for scientists analysing the two human genome sequences published by Celera and a publicly financed consortium in February.

Decoding a genome and then working out what the sequence means is a very complex task and at present it is impossible to work entirely without errors.

A team of biological information specialists at Stanford University in California, US, compared existing data on over 1,000 fruit fly proteins with protein structures predicted by Celera using their genome data.

They found that around a quarter of predictions exactly matched the existing samples and roughly another quarter was at least 99% in agreement.

'A bit rushed'

The leader of the Stanford team, Samuel Karlin, told Reuters news agency that he believed that over half of Celera's predictions disagreed with previously known information.

Front cover: Publication of the genome was a major event

"I think the publication was a bit rushed," he said.

Celera dismisses the study. "It is not a scientific paper and the Drosophila researchers who have continued the annotation do not agree with [the] interpretation of this brief commentary," Celera's Heather Kowalski told BBC News Online, referring to the fact that the Stanford team's article in the journal Nature was a brief communication rather than a full paper.

The researchers to whom she refers include Gerry Rubin of the Howard Hughes Medical Institute in Berkeley.

Taken as a compliment

He was co-author on Celera head Craig Venter's paper on the fruit fly genome that was published in the journal Science in March 2000. Dr Rubin told New Scientist magazine that the reported 50% success rate was a compliment rather than a criticism.

It is not a scientific paper.

Heather Kowalski, Celera Genomics

Work is continuing on interpreting the fruit fly genome and an update in 2002 should improve accuracy to 75%, he told the magazine.

Biologists often use Drosophila as a model organism to try out theories or techniques before applying them to more complex animals, such as humans.

There was, therefore, great interest when Celera published the Drosophila genome last year.

Annotation process

Scientists began examining the sequence and comparing it with what they already knew about the fruit fly's genes.

By itself, a genome sequence is nothing but a long series of letters describing the order of an organism's DNA.

For human beings, this sequence is around three billion letters long, and for Drosophila, it is around 1.8 billion.

A genome sequence becomes more interesting when it is broken down into individual genes.

Computer assistance

Part of this annotation process involved Celera using powerful computers to predict which bits of Drosophila's genome corresponded to which genes.

The task is all the more difficult because each single gene is usually split up across the genome, and much of the genome is apparently redundant.

It is a bit like trying to put together a jigsaw puzzle when the pieces have been torn up and mixed together with pieces of other puzzles.

Once scientists have an idea of where the genes are located, they can predict which proteins these genes are there to produce.

Plea for caution

Stanford's Samuel Karlin and his colleagues say that the uncertainties they highlighted are an indication that researchers should deal cautiously with Celera's data.

"Proteomic studies using the present Drosophila genome sequence have significant limitations and the same will be true of the human genome," they write.

They recommend that researchers using Celera's predictions attempt to replicate Celera's findings and confirm them by experiment.

Whatever the merits of either side's case, the row makes it clear that turning the blueprint of life into applicable knowledge remains a tough and controversial process.