Mad Prime

Someone made a game from this image I made! So cool!!

From Nature News:

When scientists released a draft of the human genome sequence six years ago, they said the data belonged to all of us -- but until now, they have been the only ones able to play with the data therein.

I call bullshit. That data has been publicly available, you can download it from NCBI's FTP site. Go ahead and play.

It makes me sad that most research is published under restrictive copyright licenses and is generally inaccessible to the public. But the fruit of the Human Genome Project *is* available. Find a different hyperbole, please don't needlessly promote the image of science as a monolithically inaccessible tax dollar sink.

In an ambitious attempt to transform our current scientific abilities, the X Prize foundation has announced the much-anticipated sequencing prize. 100 human genomes in 10 days is a hard task indeed, but harder yet? Doing it left-handed!

DNA is, of course, a right-handed helix. I was alerted to this invasion from the mirror universe by the Left Handed DNA Hall of Fame, where you can find a collection of many entertaining mistakes.

I know it's nit-picking, but I found this photo in Science magazine reporting on a prize specifically meant for DNA technology, so I find the X Prize Foundation's error especially deplorable. For them to get that sort of thing wrong... well, it hints at a disconnect with the actual science, obliviously taking artistic liberties in the pursuit of some media attention.
(Science 13 October 2006: Vol. 314. no. 5797, p. 232)

It seems to me that the prize conditions also reflect hype and media-pandering rather than a sincere understanding of efforts to improve sequencing technology. 10 million for 100 genomes in 10 days? Why is speed important? Why not encourage low cost sequencing? There was no cost limit announced; a company could conceivably spend much more than 10 million merely to get the prestige of the prize. That is, after all, what happened with the space prize.

If you were getting your genome sequenced, which would you rather buy: a one day wait for $100,000? Or a three month wait for $1,000? You've lived with those genes for decades, I doubt you're eager to spend a lot more money just to find out a little sooner. The real future is in cheap sequencing, not fast sequencing.

---

PS - Yeah, I know zDNA is left-handed, but you can't honestly think that this artist was intending to represent that.

PPS - What drew my eye to the picture initially were other aspects -- even flipped, this is a terrible representation of DNA. There should be only 10 bases for every turn of the double-helix (I see about 20 here). Also, the two helices are evenly spaced; they should be closer to each other so they look more like a pair twisting around (as in the cartoon), thereby forming the "major" and "minor" grooves of DNA. Lastly, the helix looks stretched-out... it isn't twisting nearly enough with respect to its width.

A couple days ago I found the most egregious error I've ever seen on wikipedia, not a graffiti issue, something that was wrong and had been wrong for a long time -- since September 15 2004, on the DNA article. A picture of the chemical structure of DNA. It was in fact a "featured pictures" candidate for September 2004; it's a little funny that all the comments about it failed to see the structure was wrong (a little sad, too).

Below is my marked-up version that points out all the errors (click it to get more resolution).

What I noticed, the immediate problem, was the base-pairing. In this picture the oxygens of guanine and cytosine were paired with each other, instead of with NH₂. It looks like the author simply rotated a DNA strand 180 degrees and lined them up, not noticing that this actually fails to orient the bases appropriately. Maybe the problem is inherent in flattening a three-dimensional structure. Maybe it's because the ribose connections of paired nucleotides are not opposite to each other, and this causes a "minor" and "major" groove in the backbones.

Anyway, I used ChemTool and GIMP to make a new picture and replaced all instances of the wrong-structure diagram with my new picture (in the articles DNA, Francis Crick, and GC content).

It took a long time, but I disapprove of people who complain about wikipedia errors without correcting them.

There are increasingly many biotechnology protocols which involve hybridizing small fragments of DNA to a large pool - methods that are sensitive to cross-hybridization. (PCR is less sensitive to this because it requires a pair of matches within a reasonable space and correct orientation, and bad matches can get weeded out by the exponential growth involved.)

So how do we check for a close match? Well, usually people run BLAST. But BLAST has been designed to look for evolutionarily close matches to a sequence -- not matches close in free energy of hybridization. It's the wrong tool for the job. It penalizes mismatches too heavily, and it treats A-T and C-G bonding as equivalent.

But I didn't want to rewrite BLAST. It'd take me months or years. Well, we can hack BLAST to fit our needs. I've figured it out and wrote up a little HOWTO guide here.

I've been getting my little science news snippets these days from Science Now news (Science Magazine, unfortunately restricted access) and Nature News (unrestricted access). I look around for other news sources, I know there's a ton out there. Today I looked at Seed magazine's news.

The top article at the time was this one: "Junk (DNA) In The Trunk".

The article's opening paragraph...

"Finding a function for the 98.5 percent of our DNA that doesn't encode for proteins - sometimes known as "junk DNA" for its jumbled, illegible arrangement - became a little less elusive last week. Geneticists from Johns Hopkins published an innovative way of using zebrafish embryos to test the purpose of non-coding human DNA sequences in the March 23rd online issue of Science Express."

Oooooh, how mysterious! We don't understand 98 percent of our DNA!

Actually, it's not.

It's not a mystery.

It's a bunch of repetitive elements, parasitic self-propagating sequences that occassionally, in frenzied bursts of self-centered replication, manage to insert copies of themselves all around the DNA. They're called transposons. 72% of our DNA is composed of retrotransposons, LINEs, and SINEs, three varieties of selfish, self-propagating junk.

This is just bad reporting. People should not propagate the mystical idea that there's vast tracts of presumably functional DNA that remain a mystery to scientists. It doesn't need a function! We're pretty sure it doesn't have much function. This sort of thing is vexing enough when it takes the form of science fiction but it's totally unacceptable in science reporting.

Of course, the reporter did not actually get any facts wrong. He simply missed the point.

This really is something interesting here. Transposable elements have been a great tool for analysis of transcriptional promotion for Drosophila, and zebrafish is an animal much more relevant. What we really care about here isn't the junk. We care about transcriptional regulatory elements, those small regions preceding genes, and maybe a few small distal elements, that determine when a gene is going to be expressed.

So, yes, there are interesting noncoding portions, but to conflate that with the 98.5% number and the term "junk DNA" is going to propagate the ignorant characterization of this stuff as being of mysterious function, when we're pretty damn sure it ain't.

---

... And, as if the world conspires to drive me apoplectic, Chris sent me a link to this article about the in silico simulation of a virus. But... what's the point? I mean, sure, it's an impressive computational feat, but what did they learn? The article failed to report on the results!

Here it is, in a quote from Nature News:

"The model also shows that the virus coat collapses without its genetic material. This suggests that, when reproducing, the virus builds its coat around the genetic material rather than inserting the genetic material into a complete coat. "We saw something that is truly revolutionary," Schulten says."

See, that's an interesting result. The LiveScience reporter missed it.

Science reporting shouldn't just be about mysteries and pretty toys. I wish science reporters didn't keep misunderstanding science and missing the point of research -- not just for the layman's sake, but mine too, because I like reading about this stuff.

DNA is fascinating to us not merely because it holds information, but because its hybridization to a partner strand means it can recognize information. Release thousands of DNA sequences into a pool, and the simple thermodynamics of hybridization means each strand will end up finding its partner. Thus the dream of DNA computing - massively parallelized by the ability of many, many small pieces to diffuse and hybridize in a solution.

The famous example is the usage of this phenomenon to solve the "travelling salesman" problem , illustrated beautifully by Larry Gonick in Discover Magazine:

Unfortunately, the field of DNA computing dries up a bit after this. It turns out there's a lot of cross-hybridization of similar sequences, and most computational problems simply can't be posed in a useful manner in a DNA hybridization context. It's sad, but the seemingly magical nature of DNA sequence and hybridization just don't translate into computation the way we'd hope.

The field of DNA computation has morphed, moved on, to the field of DNA nanotechnology. While sequence hybridization can't scale to practical computation, it can be used to create self-assembling structures. Hence the rather cute (if useless) structure gracing the cover of this week's Nature...

So here's some pics. There's a guy in my lab that works on this stuff, he emailed the article to us and we've been ogling the pretty structures woven from DNA, created by Paul Rothemund at Caltech.

It still isn't useful for anything yet. Well, that guy in my lab, he does have something useful. It's not published yet. But it's useful. The funny thing about it is that it's entirely structural, the specific nature of DNA sequence isn't used at all. In other words, someone said "we need something of this shape", and DNA just happens to be material that shape is made of.

It's a strange change from the fanciful computational nature of DNA sequence to using it as simple material for nanostructures. I think it's a practical shift of focus. Making little smiley faces does feel like the latest in a long line of useless toy constructs, and a 2D grid limits applications quite a bit, but 3D constructions are possible (if a bit wiggly). Maybe people will realize structures they can use now that the tool exists to make them.

---

PS - Larry Gonick makes excellent science-based comics. I own his Cartoon History of the Universe. I think his cartoons are awesome.

PPS - I admit, half the reason I'm posting this is because smiley faces made of DNA are so damn cute. Albeit totally useless. :-)