How we're using AI to discover new antibiotics | Jim Collins

譯者: Lilian Chiu 審譯者: Amanda Chu

So how are we going to beat this novel coronavirus?

我們要如何打敗 這種新型冠狀病毒？

By using our best tools:

用我們最好的工具：

our science and our technology.

我們的科學和技術。

In my lab, we're using the tools of artificial intelligence

在我的實驗室裡，我們用 人工智慧和合成生物學的工具，

and synthetic biology

to speed up the fight against this pandemic.

來加速對抗這場全球大流行。

Our work was originally designed

我們的研究原本是設計來 處理抗生素抗藥性危機。

to tackle the antibiotic resistance crisis.

Our project seeks to harness the power of machine learning

我們的專案計畫旨在 利用機器的學習能力

to replenish our antibiotic arsenal

來補足我們的抗生素武器庫，

and avoid a globally devastating postantibiotic era.

避免進入會重挫 全球的後抗生素時代。

Importantly, the same technology can be used

重要的是，同樣的技術能被用來 尋找抗病毒化合物，

to search for antiviral compounds

that could help us fight the current pandemic.

協助我們對抗目前的大流行。

Machine learning is turning the traditional model of drug discovery

機器學習完全顛覆了

傳統藥物研究發現的模式。

on its head.

With this approach,

有了這種方法，

instead of painstakingly testing thousands of existing molecules

就不用在實驗室裡辛苦地將數千種

one by one in a lab

既有的分子一種一種 測試過，看哪個有效，

for their effectiveness,

we can train a computer to explore the exponentially larger space

我們的做法是訓練電腦 在看似無盡的可能性中，

of essentially all possible molecules that could be synthesized,

來搜尋所有可能合成的分子，

and thus, instead of looking for a needle in a haystack,

因此，我們不是在大海中撈針，

we can use the giant magnet of computing power

而是運用電腦的強大計算能力，

to find many needles in multiple haystacks simultaneously.

像大磁鐵般同時 在好幾片大海中找到許多針。

We've already had some early success.

我們已經有一些初步的成功。

Recently, we used machine learning to discover new antibiotics

最近，我們用機器學習 發現了新的抗生素，

that can help us fight off the bacterial infections

能協助我們對抗 在新冠病毒感染時，

that can occur alongside SARS-CoV-2 infections.

會伴隨發生的細菌性感染。

Two months ago, TED's Audacious Project approved funding for us

兩個月前，TED 的「大膽計畫」 核准了我們的資金，

to massively scale up our work

讓我們能把研究規模擴大許多，

with the goal of discovering seven new classes of antibiotics

目標是要發現七種新型抗生素

against seven of the world's deadly bacterial pathogens

來對抗世界上最致命的 七種細菌病原體，

over the next seven years.

且要在接下來的七年內完成。

For context:

進一步說明：

the number of new class of antibiotics

過去三十年間發現的新型抗生素

that have been discovered over the last three decades is zero.

數目是零。

雖然我們是為了中期的未來 而在尋找新型抗生素，

While the quest for new antibiotics is for our medium-term future,

the novel coronavirus poses an immediate deadly threat,

但新冠病毒帶來了立即的致命威脅，

and I'm excited to share that we think we can use the same technology

我很興奮想跟各位分享， 我們認為同樣的技術

to search for therapeutics to fight this virus.

可以用來尋找 對抗這種病毒的治療法。

So how are we going to do it?

所以我們要怎麼做？

Well, we're creating a compound training library

我們正在建立 一個化合物訓練資料庫，

and with collaborators applying these molecules to SARS-CoV-2-infected cells

合作夥伴會將這些分子應用到 感染新冠病毒的細胞上，

to see which of them exhibit effective activity.

看看哪一種呈現有效的活動。

These data will be use to train a machine learning model

這些資料會被用來 訓練機器學習，

that will be applied to an in silico library of over a billion molecules

這些會學習的機器模型

會應用在有十億種分子的 生物電腦模擬資料庫上，

to search for potential novel antiviral compounds.

目的是尋找可能的 新型抗病毒化合物。

We will synthesize and test the top predictions

我們會針對模擬結果最有抗病毒 可能性的化合物進行合成和測試，

and advance the most promising candidates into the clinic.

並把測試結果最理想的化合物 做進一步臨床試驗。

Sound too good to be true?

聽起來太棒，不像真的？

Well, it shouldn't.

不要懷疑。

The Antibiotics AI Project is founded on our proof of concept research

我們是根據概念研究的證據 建立抗生素人工智慧計畫，

that led to the discovery of a novel broad-spectrum antibiotic

新型廣效抗生素「嗜鹽古菌素」

called halicin.

就是因此被發現的。

Halicin has potent antibacterial activity

嗜鹽古菌素有強力的抗菌活動，

against almost all antibiotic-resistant bacterial pathogens,

能對抗幾乎所有具有 抗藥性的細菌病原體，

including untreatable panresistant infections.

包括無法治療的全抗藥性感染。

Importantly, in contrast to current antibiotics,

重要的是，嗜鹽古菌素 與目前的抗生素不同，

the frequency at which bacteria develop resistance against halicin

細菌對它產生的抗藥性頻率

is remarkably low.

低得驚人。

We tested the ability of bacteria to evolve resistance against halicin

我們在實驗室中測試了 細菌對嗜鹽古菌素

及環丙沙星演化出抗藥性的能力。

as well as Cipro in the lab.

In the case of Cipro,

使用環丙沙星一天後，

after just one day, we saw resistance.

我們就看到抗藥性了。

In the case of halicin,

而使用嗜鹽古菌素一天後，

after one day, we didn't see any resistance.

我們沒看到任何抗藥性。

Amazingly, after even 30 days,

驚人的是，甚至在三十天後，

we didn't see any resistance against halicin.

我們也沒看對嗜鹽古菌素 產生任何抗藥性。

In this pilot project, we first tested roughly 2,500 compounds against E. coli.

在此試行計畫中，

我們先測試了大腸桿菌 對大約兩千五百種化合物的反應。

This training set included known antibiotics,

測試的化合物中包含已知的抗生素，

such as Cipro and penicillin,

比如環丙沙星及盤尼西林， 以及許多非抗生素的藥物。

as well as many drugs that are not antibiotics.

These data we used to train a model

我們用這些數據來訓練模型

to learn molecular features associated with antibacterial activity.

學習與抗菌活動相關的 各種分子特性。

We then applied this model to a drug-repurposing library

接著，把這個模型應用到 現有非現有抗生素藥物的資料庫，

consisting of several thousand molecules

資料庫中有數千種分子，

and asked the model to identify molecules

我們用模型來辨識

並預測會具有抗菌特性的分子，

that are predicted to have antibacterial properties

but don't look like existing antibiotics.

且和既有的抗生素不同。

Interestingly, only one molecule in that library fit these criteria,

有趣的是，在那資料庫中 只有一種分子符合這些條件，

and that molecule turned out to be halicin.

就是嗜鹽古菌素。

Given that halicin does not look like any existing antibiotic,

既然嗜鹽古菌素 與既有的抗生素並不相似，

it would have been impossible for a human, including an antibiotic expert,

因此連同抗生素專家包括在內，

to identify halicin in this manner.

無人能夠依相同的概念 以人工的方式找到嗜鹽古菌素。

Imagine now what we could do with this technology

想想看，在對抗新冠病毒上，

against SARS-CoV-2.

我們能用這種技術做甚麼？

And that's not all.

還不只如此。

We're also using the tools of synthetic biology,

我們也使用合成生物學的工具，

tinkering with DNA and other cellular machinery,

去修改 DNA 及細胞的 其他功能性構造，

to serve human purposes like combating COVID-19,

以達成人類的目標， 像是對抗新冠病毒。

and of note, we are working to develop a protective mask

重要的是，我們正在 開發一種保護性口罩，

that can also serve as a rapid diagnostic test.

它同時也具有快速診斷的檢測功能。

So how does that work?

原理是什麼？

Well, we recently showed

最近我們已證明可以從活細胞中 把功能性的構造取出，

that you can take the cellular machinery out of a living cell

and freeze-dry it along with RNA sensors onto paper

把它和 RNA 感測器 在紙上一起冷凍乾燥，

in order to create low-cost diagnostics for Ebola and Zika.

為伊波拉和茲卡病毒 創造出低成本的診斷方式。

The sensors are activated when they're rehydrated by a patient sample

感測器從病人的樣本 吸收到水份後就會啟動，

that could consist of blood or saliva, for example.

樣本可能是血液或唾液。

It turns out, this technology is not limited to paper

結果發現，這種技術 並不只限於用在紙上，

and can be applied to other materials, including cloth.

還能用在其他材料上，如布類。

For the COVID-19 pandemic,

針對新冠病毒大流行，

we're designing RNA sensors to detect the virus

我們正在設計能偵測 這種病毒的 RNA 感測器，

and freeze-drying these along with the needed cellular machinery

並將這些感測器和所需的 細胞功能性構造一起冷凍乾燥，

into the fabric of a face mask,

放進口罩的布料中，

where the simple act of breathing,

讓呼吸這般簡單的動作

along with the water vapor that comes with it,

再加上呼吸中伴隨的水氣，

can activate the test.

就能活化試劑，啟動檢測的功能。

Thus, if a patient is infected with SARS-CoV-2,

因此，如果病人帶有新冠病毒，

the mask will produce a fluorescent signal

口罩就會產生螢光信號，

that could be detected by a simple, inexpensive handheld device.

用一種經濟又簡單的手持裝置 就可以偵測到這種信號。

In one or two hours, a patient could thus be diagnosed

一兩個小時內就能安全、準確地

從遠端診斷出病人是否已感染。

safely, remotely and accurately.

We're also using synthetic biology

我們也在用合成生物學

to design a candidate vaccine for COVID-19.

設計可能對抗新冠病毒的疫苗。

We are repurposing the BCG vaccine,

我們正在改變卡介苗疫苗的用途，

which had been used against TB for almost a century.

近一世紀來，這種疫苗 一直被用來對抗結核病。

It's a live attenuated vaccine,

它是種減毒後的活體病原疫苗， 我們正在修改它，

and we're engineering it to express SARS-CoV-2 antigens,

讓它表現出新冠病毒抗原的特性，

which should trigger the production of protective antibodies

此抗原會觸發免疫系統

by the immune system.

產生保護性抗體。

Importantly, BCG is massively scalable

重要的是，卡介苗有很高的擴展性，

and has a safety profile that's among the best of any reported vaccine.

在各已知疫苗中， 它的安全性是頂級的。

With the tools of synthetic biology and artificial intelligence,

有了合成生物學 和人工智慧這些工具，

we can win the fight against this novel coronavirus.

我們可以打贏新冠病毒。

This work is in its very early stages, but the promise is real.

這項研究還在最初期的階段， 但前景確實看好。

Science and technology can give us an important advantage

在人類智慧和超級細菌 基因的對戰當中，

in the battle of human wits versus the genes of superbugs,

科技能帶給我們很重要的優勢。

a battle we can win.

我們能打贏這場仗。

Thank you.

謝謝。