"I Was Wrong": Some Refreshing Candor

One of science's saving graces is that a fair number of scientists will publicly admit that they are wrong (and then there's Marc Hauser*...). Last week, at the Human Microbiome Project meeting, Jonathan Eisen gave a talk about the GEBA project which is an effort to sequence the genomes of a diverse group of bacteria to create a bacterial genomic encyclopedia. At one point during his talk, Eisen mentioned that originally all of the genomes in the project were to be finished, although that standard has been relaxed. Eisen then noted that with the new sequencing technologies, it's feasible to sequence 1,000 genomes.

So that led to some guy named Mike asking a question about this.

Before I get to the question, it's important to explain what finishing means. When we sequence a genome--and a bacterial genome can be thought of as a circle of DNA (A, T, C, G, etc.)--we don't actually sequence the entire circle. We sequence most of it, and have large chunks known as scaffolds. For technically reasons, we can't sequence across these gaps--these 'fragmented' genomes are known as draft genomes. Closing these gaps is known as finishing (i.e., there is one scaffold for the entire genome). Unlike producing draft genomes, it is not a 'production' process. If you want to know how long it takes to finish a genome, well, we'll know when it's finished. Sometimes it takes only a few days, but other times, it can take weeks, or months. It's expensive, and requires lots of people and resources.

So back to that guy named Mike. That asshole asked Eisen (and I'm paraphrasing here), if you plan on sequencing thousands of genomes, finishing doesn't seem feasible for most of them. How important is a closed, finished genome for much of what GEBA is trying to do (e.g., discover new genes)? Eisen responded (paraphrasing again) that finishing wasn't really essential**, even though he, in older papers, argued that it was needed.

Then Eisen said, "I was wrong."

That's rarely spoken in public.

Kudos to Eisen for being honest.

*Admitting you're wrong after you get caught doesn't count.

**If you have a crappy draft with lots of scaffolds, some finishing can help to ensure that you don't have contaminant scaffolds (i.e., scaffolds that are from a different organism).

More like this

Shotgun sequencing refers to the process whereby a genome is sequenced and assembled with no prior information regarding the genomic location of any of the DNA we sequence. There are quite a few steps that you have to go through before you have an assembled genome sequence. We're going to cover…
A couple of weeks ago I attended the Human Microbiome Research Conference. At that meeting, one talk by Bruce Birren (and covered by Jonathan Eisen) mentioned something that was completely overlooked by the attendees. Now, I don't blame them, since what Birren mentioned was about bacterial…
Since I'm on-route to a Human Microbiome Project meeting (uncharacteristically, it's being held in a climate-friendly location--Houston; last year, it was held in Boston. In January.), reviewing this paper about the GEBA project, the Genomic Encyclopedia of Bacteria and Archaea, seemed appropriate…
Over at The Tree of Life, Jonathan Eisen asks: What do people think are the potential benefits that could come from finishing? For those who don't know what genome finishing is, I'll let Eisen give the short summary: Finishing: Using any combination of laboratory, computational and other analyses…

I've been wrong more times than I can count. This is not a problem: it's "the scientific method". By being wrong (and discovering why), we get closer to being right.

Thanks for the props. If you noticed the brief pause in my answer, that was a bit of pain. But I note three things:

(1) The GEBA project includes archaea. They are the sad step children of bacteria much of the time. Please don't leave them out.

(2) My answer was relating to this project. I am pretty convinced, as I answered, that most of the benefits from sampling from across the tree of life will come without completely finished genomes. And thus, the paper we wrote a few years ago, was, I guess wrong. But I did say one other thing in my answer, which was, that if you have a different goal/scientific question, finishing may be needed.

(3) The contamination issue you mention above is really important.

"Admitting you're wrong after you get caught doesn't count."

Sometimes you don't know you are wrong until somebody else points it out.

I think I know why these troublesome "scaffolds" are a problem during sequencing, but nobody seems to be interested. Simply put, chromosomal DNA is neither a simple circle nor a simple linear structure and I have evidence to suggest otherwise.

By Frank Abernathy (not verified) on 07 Sep 2010 #permalink

@2
"...if you have a different goal/scientific question, finishing may be needed."
I wish more of the genome projects would produce finished genome sequences. There are some 13 genome sequences for one of the species I study. Only one is finished, and many people say that sequence is not correctly assembled. If I want to look at horizontal gene transfer relative to chromosomal location or orientation, I can't do it if all I have is a bunch of contigs.
I am very glad they are doing the human microbiome, but I hope they will put forth the effort to finish more genomes. There may be a lot of researchers out there with "a different goal/scientific question" and the finished genomes may be useful for other scientific questions that no one has yet posed.

"Admitting you're wrong after you get caught doesn't count."

Sometimes you don't know you are wrong until somebody else points it out.

Hauser did...

By theshortearedowl (not verified) on 10 Sep 2010 #permalink