But this white one is based solely on my DNA.

但這個白色的是完全基於我的DNA。

That's not what I expected at all.

這與我的預期完全不同。

It's not an exact match, to say the least.

至少可以說，這不是一個完全的匹配。

But forensic scientists are tirelessly

但是，法醫科學家們正在孜孜不倦地

improving their ability to identify people

提高他們識別人的能力

without ever seeing their faces.

沒有看到他們的臉。

Why?

為什麼？

In criminal cases involving unidentified remains,

在涉及身份不明的遺體的刑事案件中。

recognition can be the last chance to develop leads

認可可以是發展線索的最後機會

after traditional methods fall short.

在傳統的方法不能發揮作用之後。

And if they're lucky,

如果他們幸運的話。

giving a victim their identity back could close the case.

把身份還給受害者就能結案。

That's what happened in 2017

這就是2017年發生的事情

when a company called Parabon

當一家名為Parabon的公司

created this facial reconstruction.

創造了這個面部重建。

The image got 26-year-old Shaquana Caldwell recognized,

該影像得到了26歲的沙誇納-考德威爾的認可。

and, in turn, caught her killer after months with no leads.

反過來，在幾個月沒有任何線索的情況下，抓住了殺害她的凶手。

Parabon combined two methods of facial reconstruction,

帕拉邦結合了兩種面部重建的方法。

forensic sculpture and DNA phenotyping.

法證雕塑和DNA表型。

Forensic sculpture uses skull remains

法證雕塑使用頭骨殘骸

and population samples to create an idea

和人口樣本來創造一個想法

of what the victim looked like.

的受害者看起來像什麼。

But this method won't ever be 100% accurate.

但這種方法永遠不會100%準確。

DNA, on the other hand,

另一方面，DNA也是如此。

could get us as close as we can to a copy machine.

可以讓我們儘可能地接近複印機。

That's how we ended up at Shriver Lab at Penn State.

這就是我們如何在賓夕法尼亞州立大學的施萊弗實驗室結束的原因。

Sarita: This is just envelopes and envelopes of hair.

薩里塔。這只是一包一包的頭髮。

Abby: In any other place,

艾比：在任何其他地方。

this cabinet full of hair would be really weird.

這個滿是頭髮的櫃子會非常奇怪。

They're collecting data about

他們正在收集有關的數據

people's hair type, facial features,

人們的頭髮類型，面部特徵。

and more to build an algorithm

和更多的內容來建立一個算法

to take my DNA

採集我的DNA

and spit out a replica of my face.

並吐出一個我的臉的複製品。

Producer: Does everyone sound like this when they do it?

製片人。每個人做的時候都是這樣的聲音嗎？

[all laughing]

[所有人都笑了]

Sarita: No! I've never heard anyone sound like this.

薩里塔。不！我從來沒有聽過有人像這樣的聲音。

I'm nervous I'm not going to recognize myself

我很緊張，因為我無法認出自己

when I see it on the screen,

當我在螢幕上看到它時。

not because it's not a good reproduction,

不是因為它不是一個好的複製品。

but because I

但因為我

have a different vision of myself than reality.

對自己有與現實不同的看法。

Mark: Ready? Abby: I'm ready.

馬克：準備好了嗎？艾比：我準備好了。

OK! That's not what I expected at all.

好吧!這與我的預期完全不同。

This final image is a composite

這張最後的圖片是一個合成圖

of my genetic info and the collected data

我的遺傳信息和收集的數據

from people who share my genetic ancestry.

來自與我有相同遺傳血統的人。

Right now, the algorithm works best,

現在，該算法的效果最好。

meaning the predicted face is most recognizable,

意味著預測的臉是最容易被識別的。

when a person looks, well,

當一個人看起來，好。

statistically average.

在統計學上是平均的。

So, you can get ancestry axes

所以，你可以得到祖先的軸心

estimated from the genotypes,

從基因型中估計。

and then we layer on top

然後我們在上面分層

the feature-specific genotypes.

的特定特徵基因型。

Abby: Mm, gotcha.

艾比：嗯，知道了。

There are some big differences

有一些很大的區別

between my actual and predicted faces.

我的實際面孔和預測的面孔之間。

Mark: Which is telling me that

馬克：這是在告訴我，

we need more information in the model.

我們在模型中需要更多的資訊。

Abby: Humans share about 99.9% of DNA,

艾比：人類共享大約99.9%的DNA。

but what makes me look like me

但什麼使我看起來像我

is in the last 0.1%.

是在最後的0.1%。

And that's thanks to single-nucleotide polymorphisms,

而這要歸功於單核苷酸多態性。

the variations in base pairs throughout our genes.

我們整個基因中的鹼基對的變化。

This is easiest to see with our eyes.

這是最容易用我們的眼睛看到的。

I have two adenine at the base pair,

我在鹼基對上有兩個腺嘌呤。

most associated with having brown or hazel eyes.

與擁有棕色或淡褐色眼睛最相關。

Which I do.

我就是這樣做的。

Now, if you take the A's out,

現在，如果你把A隊去掉。

and you replace them with two guanine,

而你用兩個鳥嘌呤取代它們。

then that likelihood shifts

那麼這種可能性就會轉移

towards blue or green.

朝向藍色或綠色。

Mark and his team are building a database

馬克和他的團隊正在建立一個數據庫

of which SNPs affect which features.

哪些SNP影響哪些特徵。

Mark: We're really at an early stage.

馬克：我們真的處於一個早期階段。

I mean, we have 200 genes

我的意思是，我們有200個基因

that have significant effects on the face.

對面部有顯著影響的。

Abby: At the moment, their formula isn't perfect.

艾比：目前，他們的配方並不完美。

A lot of traits are controlled by multiple SNPs,

很多性狀是由多個SNP控制的。

so it isn't a simple one-to-one connection.

所以它不是一個簡單的一對一的連接。

You know, the distinctive features

你知道，與眾不同的特徵

that we really rely on to identify each other

我們真正賴以識別彼此的是

are not necessarily coming through.

不一定能通過。

Abby: Small details are missing,

艾比：小細節缺失。

like my cheek mole.

像我臉頰上的痣。

And my nose is a bit pointier in reality.

而我的鼻子在現實中是比較尖的。

I don't usually think I look like my dad,

我通常不認為我看起來像我父親。

but without the mapping,

但沒有映射。

the texture mapping, I think,

我認為是紋理映射的問題。

this looks a lot like him.

這看起來很像他。

Then those are probably genes with big effects,

那麼這些可能是有大影響的基因。

because, you know, those differences that you see

因為，你知道，你看到的那些差異

that make somebody distinctive

具備獨特個性的人

often stand out in their siblings.

往往在他們的兄弟姐妹中脫穎而出。

Like you said, your face looks like your dad.

就像你說的，你的臉看起來像你爸爸。

Abby: Plus, SNPs are associated with different traits,

艾比：另外，SNP與不同的性狀有關。

not definitive proof of them.

並非是它們的確切證據。

Like, I have all the right SNPs

比如，我有所有正確的SNPs

for not having a cleft chin,

因為沒有裂開的下巴。

but tell that to all of my bullies

但請告訴我所有的欺負我的人

who called me butt-chin growing up.

在我成長的過程中，他叫我屁精。

With so much variation,

有這麼多的變化。

to get this closer to this,

以使其更接近於此。

Mark and his team need data from as many people as possible.

馬克和他的團隊需要儘可能多的人提供數據。

And so if you already have our DNA,

是以，如果你已經有了我們的DNA。

what are we getting here?

我們在這裡得到什麼？

Sarita: In order to inform our research,

薩里塔：為了為我們的研究提供資訊。

we need to take measurements and data on many phenotypes.

我們需要對許多表型進行測量和數據。

The idea is that, hopefully,

我們的想法是，希望如此。

by collecting all of this data

通過收集所有這些數據

and associating it with what you really look like,

並將其與你的真實面貌聯繫起來。

we can then use that to inform

然後我們可以用它來通知

an algorithm to predict on someone

預測某人的算法

that we don't know this information about.

我們不知道這些資訊的人。

Abby: So I decided to offer up my data to the algorithm.

艾比：所以我決定把我的數據提供給算法。

All right, snip away.

好了，剪掉吧。

Hair can make or break a case,

頭髮可以成就或破壞一個案件。

but the most common form of forensic hair analysis

但最常見的法醫毛髮分析形式是

isn't scientific.

是不科學的。

Sarita: The most common forensic form of hair analysis

薩里塔最常見的法醫毛髮分析形式

often uses racial categories to categorize hair.

經常使用種族類別來對頭髮進行分類。

And so, in this way,

是以，以這種方式。

we're able to study hair more objectively.

我們能夠更客觀地研究頭髮。

So we're going to embed your hair

所以我們要嵌入你的頭髮

in a low-melt-point plastic

在一個低熔點的塑膠中

and then cut that embedded chip in half,

然後將該嵌入的芯片切成兩半。

and that'll allow us to look down the shaft of the hair.

這將使我們能夠順著頭髮的軸線看下去。

And so we call that cross-sectional.

是以，我們把這稱為橫斷面。

And we can look at the elliptical nature of your hair,

而且我們可以看看你的頭髮的橢圓性。

whether it's more circular, the thickness,

是否更圓，厚度。

see if you have a medulla,

看看你是否有一個髓部。

because some people don't have a medulla in their hair.

因為有些人的頭髮中沒有髓部。

Abby: What is a medulla?

艾比：什麼是髓部？

Sarita: The medulla is the center portion of the hair.

薩里塔。髓部是頭髮的中心部分。

You can kind of think of it as like

你可以把它想成是

the bone marrow of the hair.

頭髮的骨髓。

This is really important forensically

這在法醫上真的很重要

because if you are a forensic scientist,

因為如果你是一個法醫學家。

you get up on the stand, you say it's an African hair type,

你站在證人席上，你說這是一個非洲人的頭髮類型。

then you're going to sway the jury into thinking

那麼你就會動搖陪審團的想法

that the perpetrator was a specific race.

肇事者是一個特定的種族。

Abby: They also sample your skin pigmentation.

艾比：他們還對你的皮膚色素進行採樣。

Javier: So, what this does is it measures

哈維爾：所以，這所做的是測量

the melanin in your skin.

你皮膚中的黑色素。

I'm going to need three parts.

我將需要三個部分。

I'm going to need this part of your arm right here,

我需要你手臂的這一部分，就在這裡。

with your palm facing up,

掌心朝上。

because it's not exposed to a lot of light.

因為它沒有暴露在大量的光線下。

Abby: Oh, yeah, I hide that. Javier: We need your forehead,

哦，是的，我隱藏了這個。哈維爾：我們需要你的額頭。

because that's exposed to a lot of light actually.

因為這實際上是暴露在大量的光線下。

And we're going to need your hair,

而我們將需要你的頭髮。

because we're going to be also looking at

因為我們也要看一下

your hair pigmentation.

你的頭髮色素沉著。

Abby: This little thing is sort of like

艾比：這個小東西有點像

what they use in makeup stores.

他們在化妝品商店使用的東西。

Javier: Yep, that's right.

哈維爾：是的，這是對的。

Abby: For weight and BMI,

艾比：對於體重和BMI。

the Tanita scale shoots a small amount of electricity

谷田衡器射出少量的電力

through your body.

通過你的身體。

Sarah: OK, we're going to electrocute you.

莎拉：好的，我們要電擊你。

Abby: OK.

艾比：好的。

Sarah: Small. Abby: What about my mic pack?

薩拉：小。艾比：我的麥克風包呢？

Sarita: It should be fine.

薩里塔：應該沒問題。

Abby: OK.

艾比：好的。

Oh, no. Sarah: You don't feel it.

哦，不。莎拉：你不覺得。

Abby: I know, I'm just nervous.

我知道，我只是緊張。

Because I'm a participant in the research,

因為我是研究的參與者。

they need to know what my actual face looks like.

他們需要知道我的真實面目是什麼樣子。

Sam: So, this is essentially our low-tech

所以，這基本上是我們的低技術含量的

zero-gravity machine.

零重力機。

[both laughing]

[兩人都笑了]

Oh, I feel it. I'm in space.

哦，我感覺到了。我在太空中。

Gravity affects how your face lies on top of your skull.

重力會影響你的臉在頭骨上的位置。

So getting photos from all angles

是以，從各個角度獲取照片

provides a more accurate final product.

提供一個更準確的最終產品。

Sam: Basically what we'll do is we'll take all the photos

基本上我們要做的是，我們將拍攝所有的照片。

from all the angles that we took,

從我們拍攝的所有角度來看。

put them in to our vector software,

把它們放到我們的矢量軟件中。

and actually stitch them together.

並實際將它們縫合起來。

So we'll get something that you can actually

是以，我們將得到一些你可以真正

revolve around on a computer screen.

圍繞著電腦屏幕旋轉。

Abby: You're going to put this up next to

艾比：你要把這個擺在

the recreation from my DNA

從我的DNA中獲得的再現

and see how they match or not?

並看看它們是否匹配？

Sam: Yeah, we're going to see how accurate

是的，我們要看看有多準確

we can predict your face.

我們可以預測你的臉。

Abby: By using known faces to check their work,

艾比：通過使用已知的面孔來檢查他們的工作。

they can hone the accuracy of the algorithm

他們可以磨練算法的準確性

when predicting unknown faces.

在預測未知面孔時。

And the more times they see SNPs correlating

而且他們看到SNPs相關的次數越多

with different traits,

具有不同的特質。

the more definitively they can connect them.

他們越是能明確地把它們聯繫起來。

We don't need to sample everybody

我們不需要對每個人進行採樣

to understand how something works.

以瞭解某物的運作方式。

We need a representative sample ethnically,

我們需要一個有代表性的種族樣本。

or population appropriate.

或適合人群。

I wouldn't take our methods and the data we've collected

我不會把我們的方法和我們所收集的數據

and go to Finland and say,

併到芬蘭去說。

OK, let's set up a big phenotyping lab here,

好吧，讓我們在這裡建立一個大的表型實驗室。

without sampling Finnish people.

沒有對芬蘭人進行抽樣調查。

Abby: There's still a long way to go

艾比：還有很長的路要走

before this technology can become mainstream.

在這項技術能夠成為主流之前。

One of my deep concerns over the forensic applications

我對法醫應用的深切關注之一是

of molecular photofitting

分子的光合作用

are the premature deployment of the methodology.

是該方法的過早部署。

It's important that investigators

重要的是，調查人員

understand what you found

瞭解你發現的情況

and will use that information that you found appropriately.

並將適當地使用你找到的這些資訊。

Otherwise, you give the investigator

否則，你給調查員

an image of what the person might look like

該人可能長什麼樣的影像

and they can say, "Well,

而他們可以說，"好吧。

that image is not the same as my defendant."

該影像與我的被告人不一樣。"

Abby: The team needs to collect more data,

艾比：團隊需要收集更多數據。

improve the computational power of the algorithm,

提高算法的計算能力。

and simply more time to get things right.

和簡單的更多時間來把事情做好。

But even then, DNA is just your blueprint,

但即使如此，DNA也只是你的藍圖。

not the final product.

而不是最終產品。

As we mentioned, SNPs rely on probability.

正如我們所提到的，SNPs依賴於概率。

Plus, there's plenty of environmental factors

此外，還有很多環境因素

that come into play.

這將會發揮作用。

We don't expect every feature is predictable.

我們並不期望每一個功能都是可預測的。

Even identical twins.

甚至是同卵雙胞胎。

Abby: It can't tell us things like a person's gender,

艾比：它不能告訴我們像一個人的性別的事情。

if they have scars, or if they dye their hair

如果他們有疤痕，或如果他們染髮

or wear makeup.

或化妝。

Some might enjoy the sun much more than the other twin,

有些人可能比另一個雙胞胎更喜歡陽光。

and you can see those kinds of differences emerging

而且你可以看到這些種類的差異出現

because of the environment.

因為環境的原因。

Abby: The key to using DNA phenotyping to solve crimes

艾比：利用DNA表型來破案的關鍵

isn't perfecting the tech.

並不是在完善這項技術。

It's understanding its limitations.

這是對其侷限性的理解。

This has been a 20-year project already,

這已經是一個20年的項目了。

and, you know, I expect will continue

而且，你知道，我預計將繼續

in the next five, 10 years.

在未來5年或10年內。

We'll have more than double our current sample size

我們將有超過目前兩倍的樣本量

and be able to say a lot more about appearance.

並能對外觀說得更多。

All right. Abby: Thank you so much.

好的。艾比：非常感謝你。

Sure.

當然。

Do I have a Medusa?

我有一個美杜莎嗎？

Medulla, sorry.

髓部，對不起。

Sarita: Oh, is that what you were trying to say?

薩里塔。哦，這就是你想說的嗎？

Abby: That's what I was trying to say.

艾比：這就是我想說的。