"Trivial" Error in Celera Fly Genome

In "one of the most petty and ridiculous issues ever in the history of science" - according to the guilty company's chief scientific officer - federal officials noticed that when Celera uploaded genetic sequences of the fruit fly to a public database, there was some human DNA mixed in. It's now been removed, and everyone seems to think this is not a very significant error. But the harsh exchange and defensive posturing on both sides underscores the edgy rivalry between the government group's slow-and-steady approach and Celera's "shotgun" approach to mapping the human genome. This story is also important because mixing human genes into a fly's genes is freaky cool - someone should make a movie about that.

Favourite "Fly Crossed With Human" Movie

[Anonymous Coward]

...definately "Tom Thumb", the claymation movie. It you haven't see it, it's a must!

Open Source Applies Here Too

[dwdyer]

Did anyone catch this quote from the Wired article?

"The advantage to making data public is that people can find your mistakes as quickly as possible," Scherer said. "By holding onto it privately no one knows and those mistakes would be propagated forward."

It seems that the principle advantage of open source software is a principle of open source information as well.

freaky cool

[snyrt]

that's totally freaky cool, they should make a movie out of that. i mean, it'd be like the tick, but the fly, or maybe it'd be like the tick's friend the moth guy dude, but that would be cool. i bet if they plugged in that bad genome to their gene mapping software, they could figure out exactly what amino acids would be human and which would be fly. that would mean that they would be able to accurately figure out if they could possibly make something like that. whoa, wait, here he comes, AAAAAAAAAAAAHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH!!!! !!!!!!!!!

A little more info, from Nature...

[Guppy]

Here's some info from the latest issue of Nature [nature.com]. Their site requires a login, so I'll just post an exerpt below. The article gives a little more detail on the story, such as the actual sizes of the sequences involved.

Drosophila genome contaminated with human sequence

[WASHINGTON] Computational biologists at the National Center for Biotechnology Information (NCBI) have discovered about 150,000 bases of human DNA mixed in with the genomic sequence of the fruitfly Drosophila compiled by Celera Genomics of Rockville, Maryland, and the Berkeley Drosophila Genome Project (BDGP).

Scientists working with the data were downplaying the significance of the contamination last week. David Lipman, NCBI director, says that the figure should be seen in the context of the approximately 180 million bases of the fly genome.

Lipman adds that the errant data were in a part of the database holding 'unassigned scaffolds', not the labelled part of known fly chromosomes. "We found no evidence of any contamination in the main body of fly sequence data," he says.

Researchers at Celera and the BDGP had previously pointed out that some of the unassigned sequence might contain foreign DNA. They have requested a correction in Science, which published the Drosophila sequence last month. Gerald Rubin, leader of the BDGP, has been quoted as describing the contamination as "trivial".

Re:It's the less obvious errors that worry scienti

[lovebyte]

What is even more worrying is that Celera has not blasted their drosophilia sequences against their own homo sapiens sequences to find potential contaminations. Or maybe they did and did not want to tell anyone!

Errors not so uncommon.

[No Such Agency]

I'm a molecular biologist who works frequently with "database homology searches" and I can tell you that there are actually quite a few errors in sequence information submitted to GenBank and other public sequence databases. Usually they are fairly minor, such as the guy in our lab whose submitted protein sequence contained letters from the following sentence of his thesis on the end! Most others are just from incorrectly read data at the lab-bench level I think. But this seems like a pretty big screw-up by most standards, one that could have been caught if they weren't in such a freakin' hurry! Oh well, at least it was caught.

It's the less obvious errors that worry scientists

[Guppy]

I'm not surprised the error was caught so quickly. One of the first things that was probably done with the Fly data was to run a sequence matching program (Like NCBI BLAST [nih.gov]), to compare it to the known portions of the human genome. The sequence matches would stand out, as they would be ~100% identical. That might occasionally happen for very short sequences, but exact matches with any large portions of the human genome would be highly suspicious.

I think the problem here (Besides egos and bad blood) is that some scientists are still suspicious of Celera's motives and methods, and that they wonder if sometime in the future Celera could (whether accidentally or on purpose) release corrupted data that would be much harder to detect.