'It started with hair.'
May 3, 2013 9:00 AM   Subscribe

Artist Heather Dewey-Hagborg uses the DNA found on things like discarded chewing gum and cigarette butts to recreate the faces of the people who left them behind.
posted by Chutzler (57 comments total) 18 users marked this as a favorite
 
That is really cool. Thanks.
posted by Elementary Penguin at 9:02 AM on May 3, 2013


I need a control sample or something!

Like, maybe if she would take a sample of her hair, run it through this program, then model herself, so I could see side by side.

Because here's the thing is I'm totally dumb about this stuff. Hollywood has totally spoiled me. What kind of data can you draw about facial structure from a DNA sample? Like, is there any chance these are remotely accurate?

I feel stupid for asking. Maybe I should go back and more closely read the article.
posted by kbanas at 9:08 AM on May 3, 2013 [4 favorites]


I somehow misread that so I thought she was recreating their faces using used gum and cigarette butts. Eew. Though I suppose that would also say something about filthy habits.
posted by biffa at 9:08 AM on May 3, 2013 [3 favorites]


The first picture in the article is a self-portrait, kbanas.
posted by Elementary Penguin at 9:08 AM on May 3, 2013 [4 favorites]


Ok, so, it says, "There is, of course, no way of knowing how accurate Dewey-Hagborg’s sculptures are—since the samples are from anonymous individuals, a direct comparison cannot be made."

So, that's true, but if she subjected herself (or some other known quantity) to the process, you could get some sort of comparison.
posted by kbanas at 9:10 AM on May 3, 2013


The first picture in the article is a self-portrait, kbanas.


Ok. I fail today. Really cool article. I apparently have lost all ability at reading comprehension and will show myself out.
posted by kbanas at 9:11 AM on May 3, 2013


A little googling around suggests that DNA facial reconstruction remains out of reach as yet for forensic purposes, but is almost certainly a near-future reality. Which makes sense, of course; identical twins tell us that facial features are highly genetically determined; the tricky part is learning to read that code.
posted by yoink at 9:15 AM on May 3, 2013


I really want to be able to send her a swab of my DNA and have her do my portrait.

That would satisfy the desire for scienciness and also be awesome!
posted by jacquilynne at 9:17 AM on May 3, 2013 [2 favorites]


This whole project is anecdotal. She's not "reconstructing faces"; she's making them up, based on "ancestry, gender, eye color, propensity to be overweight and other traits". The science here is telling her "brown-eyed Italian male"; she's making up the rest.
posted by Fnarf at 9:20 AM on May 3, 2013 [12 favorites]


I don't disagree, kalessin. I would be interested in more rigorous details of her process, since it sounded like the process is based on her own research into correlations between DNS markers and facial structure.
posted by Elementary Penguin at 9:20 AM on May 3, 2013


This whole project is anecdotal. She's not "reconstructing faces"; she's making them up, based on "ancestry, gender, eye color, propensity to be overweight and other traits". The science here is telling her "brown-eyed Italian male"; she's making up the rest.


Right. This is an art project, with a little bit of science thrown in to get it started.
posted by kbanas at 9:21 AM on May 3, 2013 [1 favorite]


That's why I eat all my cigerette butts.

Next up: reconstructing other body parts using pubes left behind at porn arcades.

Yeah! I use to tell gay men that unwantedly hit on me, who asked "what are you into?", I'd always respond [deadpan] Collecting pubes I find the the bathroom.
It worked, until one day when it didn't...
posted by QueerAngel28 at 9:29 AM on May 3, 2013 [7 favorites]


From the article:
Certainly, there are limitations to what is known about how genes are linked to specific facial features.”We are really just starting to learn about that information,” says Dewey-Hagborg. The artist has no way, for instance, to tell the age of a person based on their DNA. “For right now, the process creates basically a 25-year-old version of the person,” she says.
...
She sends the mitochondrial DNA to a lab to get sequenced, and the lab returns about 400 base pair sequences of guanine, adenine, thymine and cytosine (G, A, T and C).

From this sequence, Dewey-Hagborg gathers information about the person’s ancestry, gender, eye color, propensity to be overweight and other traits related to facial morphology, such as the space between one’s eyes. “I have a list of about 40 or 50 different traits that I have either successfully analyzed or I am in the process of working on right now,” she says.

Dewey-Hagborg then enters these parameters into a computer program to create a 3D model of the person’s face.” Ancestry gives you most of the generic picture of what someone is going to tend to look like. Then, the other traits point towards modifications on that kind of generic portrait,” she explains. The artist ultimately sends a file of the 3D model to a 3D printer on the campus of her alma mater, New York University, so that it can be transformed into sculpture.
There's significant guess-work involved, but 40-50 traits isn't as vague as "brown-eyed Italian male".
posted by filthy light thief at 9:33 AM on May 3, 2013 [5 favorites]


I kind of wish she would pair the portrait with the remainder of the trash she found, and a time and location of where she found the trash.

Really, I wish this included some public shaming about leaving chewing gum and cigarette butts around.
posted by filthy light thief at 9:34 AM on May 3, 2013 [6 favorites]


The article is complete BS. She probably gets some ethnicity information and then makes everything else up. The portrait of herself proves nothing, since she already knows what she herself looks like. It's not a blind test.

on preview

...40-50 traits isn't as vague as "brown-eyed Italian male".

That would depend on the traits. "Italian" isn't a trait that's encoded in your genes. Hair color, skin color, etc are. We recognize certain trait clusters as "Italian".

And that's also assuming she isn't just BSing. Where's the double-blind study? Granted, it's an art project. But the article, with significant help from her, makes it sound like science. Either get skeptical or make it clear that this isn't science.
posted by DU at 9:36 AM on May 3, 2013 [4 favorites]


DU, good point on the traits.
posted by filthy light thief at 9:37 AM on May 3, 2013


It's even hard to tell if her portrait looks much like her and to what extent it was done "blind." The method described is a little bit cheesy, depending heavily on racially stereo typical facial features and a few tidbits of known mutations related to appearance. The extent to which this is artful is questionable, but it is definitely not scientific. [Or, on edit, what DU said.]
posted by Mental Wimp at 9:38 AM on May 3, 2013


I use comments people leave laying around internet discussion boards to reconstruct how well they can read and understand news stories. There's a little bit of art to it, but the few instances I had tried to validate have proved fairly accurate.
posted by benito.strauss at 9:38 AM on May 3, 2013 [14 favorites]


I don't understand how she can get 40-50 SNPs from 400 bp of mitochondrial DNA.
posted by Cold Lurkey at 9:39 AM on May 3, 2013 [1 favorite]


Does she or doesn't she? Only her mitochondria know for sure.
posted by Atom Eyes at 9:51 AM on May 3, 2013 [2 favorites]


Can I use dna to see who didn't pick up after their dog?
posted by Obscure Reference at 10:02 AM on May 3, 2013


If they actually could pull an accurate portrait from a cigarette butt, I'd be all over it just for the public shaming of those who think it's A-OK to just chuck those things out their car windows.
posted by jquinby at 10:02 AM on May 3, 2013 [1 favorite]


This is totally real. Angela does way more intricate stuff than this on Bones.

TV counts as real, right?
posted by rikschell at 10:14 AM on May 3, 2013 [3 favorites]


She sends the mitochondrial DNA to a lab to get sequenced, and the lab returns about 400 base pair sequences of guanine, adenine, thymine and cytosine (G, A, T and C). - article

I don't understand how she can get 40-50 SNPs from 400 bp of mitochondrial DNA. - Cold Lurkey

Plus mitochondrial DNA doesn't tell you much about the individual it came from. It's inherited through the mother directly - you have absolutely no mitochondria from your father. So it tells you about your mother's lineage to an extent.

Note: Unless I am misremembering how mitochondria are inherited or there's newer data than I'm familiar with.
posted by maryr at 10:30 AM on May 3, 2013


Obscure Reference: "Can I use dna to see who didn't pick up after their dog?"

There's actually a business that does that.
posted by exogenous at 10:30 AM on May 3, 2013 [1 favorite]


Not only creepy but bullshit too!

But I'll bet our genes pretend to say we were all descended from a Cherokee princess too.
posted by spitbull at 10:30 AM on May 3, 2013 [2 favorites]


To follow up my comment above, that's why they've used mitochondrial DNA to find whatever small number of "Eves". For "Adams" you use the Y chromosome, inherited solely from the father.
posted by maryr at 10:31 AM on May 3, 2013


I have visions of receiving $20 fines in the mail with the messaging, "Your DNA was extracted from the wad of gum located on the poster of Tom Cruise at the Xth Street subway station. You are fined $20 for littering and defacing a poster of the President of the United States."


Okay, so I decided to make the reader wonder, "WHICH IS CREEPIER?"


Gattaca is how far away?
posted by Atreides at 10:38 AM on May 3, 2013 [3 favorites]


Yeah, the mitochondrial thing is a major red flag. At best, she's looking at some very general markers of maternal ancestry. I mean there are only 37 genes on the mitochondrial genome (as opposed to 30K on the chromosomes) and most of them are involved in a very specific part of energy metabolism. If there's any information there, it's going to be about roughly what part of the world the subject's mother was from - and even then it's not going to be a particularly reliable signal. Maybe the article isn't accurately describing what she's doing (did they mean she gets back multiple 400bp fragments?), but as is, it definitely doesn't make a lot of sense.
posted by en forme de poire at 10:43 AM on May 3, 2013 [3 favorites]


(And President Cruise is way creepier.)
posted by en forme de poire at 10:43 AM on May 3, 2013


Next up: reconstructing other body parts using pubes left behind at porn arcades.

I'm thinking the resulting creation would look like it was assembled from a spare-parts bin, with a few extra limbs.
posted by zippy at 10:45 AM on May 3, 2013


Her self portrait didn't strike me as very accurate, other than, you know...caucasian and female...
posted by stenseng at 11:02 AM on May 3, 2013


I predict a future where this technique is regularly misused by law enforcement.
posted by Chinese Jet Pilot at 11:17 AM on May 3, 2013


Creepy. And I won't be spitting my gum out the window anymore.
posted by yoga at 11:28 AM on May 3, 2013


Her self portrait didn't strike me as very accurate, other than, you know...caucasian and female...

She had the opportunity to make it look like whatever she wanted it to, within the parameters of those 40 or 50 features specified by the DNA. You'd expect it to look almost exactly like her... or at least, almost exactly like how she thinks she looks.

I'd really like to see her try to sculpt somebody blind and then compare the sculpture to the person.
posted by painquale at 11:38 AM on May 3, 2013


Also, she has to be looking beyond the mitochondrial genome if she's inferring the subject's sex from these. Mitochondria are the same in men and women.
posted by en forme de poire at 11:52 AM on May 3, 2013


ball of fingers
posted by exlotuseater at 11:55 AM on May 3, 2013


This is totally fake, by the way. I mean, you could maybe get hair color and eye color if you had $100k per sample. Ancestry would be much easier but it would be very general -- "probably Caucasian," not "Italian." And mtDNA doesn't even make sense for this purpose.

Maybe in 20 years we'll have this but not yet.
posted by miyabo at 11:59 AM on May 3, 2013


For those that think that this is something real: she uses 400 base pairs, out of 3.2 billion that the human genome has. Do you still think it's possible?
Let's be very optimistic and say each base pair is 2 bit, so 800 bit. If you copypaste this comment into a text file and save it, it will be bigger than that. Still think it's enough to reconstruct a human face?
posted by dhoe at 11:59 AM on May 3, 2013


If you copypaste this comment into a text file and save it, it will be bigger than that. Still think it's enough to reconstruct a human face?

I fed this text into the analysis: result, though I don't know if it's accurate.
posted by zippy at 12:03 PM on May 3, 2013 [4 favorites]


Yeah, this is completely fake.
posted by Justinian at 12:15 PM on May 3, 2013


You can infer a small portion of your maternal ancestry from mitochondrial DNA's (mtDNA's) hypervariable regions, which I assume is what the article refers to. As maryr and others have said, it's passed down from mother to children. So the tiny region on my mtDNA tells me something about where my mother's mother's mother's mother's mother'smother's etc. most likely came from. This is only one of the many ancestors that have contributed DNA to me. It tells you nothing really about phenotypic (physical) traits, except for maybe broad generalizations in the form of, "people from ___ typically have ___ colored eyes." Though the science for genetic facial reconstruction is nowhere near mature, you could do better with something like 23andme's genotyping service. Like you could be fairly certain if the DNA source (which is quite possibly contaminated) has wet or dry earwax. Otherwise this is mostly nonsense.
posted by bergeycm at 12:27 PM on May 3, 2013 [1 favorite]


I actually think this is a cool concept, and there's something compelling about both the story and the 3D-printed face masks - but yeah, scientifically, very unlikely to be literally true.
posted by en forme de poire at 2:22 PM on May 3, 2013


So, neither scientific nor artistic.
posted by anazgnos at 3:05 PM on May 3, 2013


It's absolutely a cool concept. So is affordable and efficient fusion power. They are equally present in the artist's work at the present time.
posted by Justinian at 3:08 PM on May 3, 2013


I had no idea DNA technology was this advanced (though many people in this thread are saying it is fake)? Do police use this technique when they collect DNA from crime scenes to get an idea of what the suspect looks like?
posted by pravit at 3:09 PM on May 3, 2013


I think its not scientific, but a fun art project. I think it would be more interesting to be able to send her a sample of your DNA (with your identifying info stripped by a third party) and then get back her artistic rendering of your face. Interesting in a fun kind of way. I would pay for that on etsy probably. It would be like the sculpture version of those DNA art prints people buy.
posted by Joh at 3:14 PM on May 3, 2013


Do police use this technique when they collect DNA from crime scenes to get an idea of what the suspect looks like?

It's not real. But I repeat myself.
posted by Justinian at 3:16 PM on May 3, 2013 [1 favorite]


(Note: I'm speaking of the facial reconstruction bit, not the sex etc bit)
posted by Justinian at 3:19 PM on May 3, 2013


I am a little ashamed of myself. I,was skeptical when I saw the link but figured "the Smithsonian magazine wouldn't print nonsense!" Too trusting, I guess.
posted by Elementary Penguin at 4:36 PM on May 3, 2013


It is interesting indeed that the article is completely buying it, and this paragraph has me scratching my head:

Hal Brown, of Delaware’s medical examiner’s office, contacted the artist recently about a cold case. For the past 20 years, he has had the remains of an unidentified woman, and he wondered if the artist might be able to make a portrait of her

So what is this, intentional misrepresentation? Or is the article part of some lame art project where they then say that people trust science too much?
posted by dhoe at 5:24 PM on May 3, 2013


To be fair, police investigators of cold cases have hired psychics. I think after 20 years you get desperate.
posted by Elementary Penguin at 5:40 PM on May 3, 2013


I've been stewing over this all day, and I think I've distilled my distaste down to 2 major points:
1. Art like this does not help bridge the divide between Science and your average Civilian. Instead, it further obscures the unknowns that Scientists contend with on a daily basis and does nothing to explain what IS known, but rather suggests that there exists a magic process of decoding that you, the Civilian, will never really understand, so don't bother your head about it. At best it's disingenuous about genomes and our current understanding of them, and at worst is simply trying to be provocative, creating a fantasy of "what if we could track you down by some scraps of DNA you dropped accidentally on the street?" What dystopian hell would we have created?
Sub point 1. Currently, we are still at an impasse regarding what we can and cannot do, ethically, medically, insurancely, with the basic data and understanding of genomes we have now. Skipping ahead of ourselves seems a bit... well, reckless.
2. So, if we assume that the Smithsonian is at least accurate with the description of her methods, (which honestly I doubt, sounds like some factchecking error, cuz seriously, who only gets 400bp reads from sanger sequencing anymore) she's really rendering a 3D form based off of differential information from a limited scrap of DNA. This isn't really different from the "MUSIC FROM DNA" nonsense or ugh, just the worst piece of art i've ever seen, which was a big old print of an absolutely run of the mill electrophoresis gel. It's at the Harris Theater in Chicago, if anyone can help me track it down, and it's utterly meaningless. That is neither science nor art, as it's just an image, without information or content for either the Civilian viewer nor the Scientist. I guess, as a geneticist, I'm more than a bit tetchy when I see people using DNA sequence as an algorithm or random number generator when I spend all my time trying to get the DNA to tell me important things about what it does in the developing organism.
Sub-point 2. (self pitying geneticist indulgence please) I attempt to be as artistic of a Scientist as I can. I went to the same undergrad as Dewey-Hagborg, which was arguable more of an art-school than a liberal arts school. I strive to make my research as creative, inventive, evokative and expressive of the world-as-it-exists as possible. I feel this approach is essential to expose all the beautiful ideas that lie, just now, beneath the surface of our understanding. If I truly believed it, I would now trot out the tired line of "SCIENCE IS ART", but I don't , so, I'll just caution the artists using science artistically to remain clear about the border between the possible, the fantastic, and the real. And then sternly remind them that using the veil of science to obscure the fact that you're just dicking around with the colors and form of a DNA gel or using genomes as random number generators is not sufficient to make it compelling.
posted by Cold Lurkey at 5:46 PM on May 3, 2013 [2 favorites]


She has a press release (PDF) on her site which sounds much saner. Nothing about mDNA or 400 bp, for starters. She takes SNPs to get "gender, ancestry, eye color, hair color, freckles, lighter or darker skin, and certain facial features like nose width and distance between eyes", and works from there.
posted by dhoe at 12:40 AM on May 4, 2013 [3 favorites]


So this is all possibly the fault of bad science reporting? That is some cold comfort.
posted by Elementary Penguin at 2:56 AM on May 4, 2013


I've been stewing over this all day

And that's why it's great art.
posted by telstar at 2:53 PM on May 4, 2013


Yeah, all bad journalism is great art.
posted by dhoe at 10:16 AM on May 5, 2013 [2 favorites]


« Older Can you brown this up for me?   |   Blelvis Newer »


This thread has been archived and is closed to new comments