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

40,414 views ・ 2020-05-26

TED


Dvaput kliknite na engleske titlove ispod za reprodukciju videozapisa.

Prevoditelj: Jasmina Sevo Recezent: Sanda L
00:12
So how are we going to beat this novel coronavirus?
0
12917
2908
Dakle, kako ćemo pobijediti ovaj novi koronavirus?
00:16
By using our best tools:
1
16317
2631
Korištenjem naših najboljih alata:
00:18
our science and our technology.
2
18972
2039
znanosti i tehnologije.
00:21
In my lab, we're using the tools of artificial intelligence
3
21594
3132
U mom laboratoriju koristimo alate umjetne inteligencije
00:24
and synthetic biology
4
24750
1579
i sintetičke biologije
00:26
to speed up the fight against this pandemic.
5
26353
3060
kako bismo ubrzali borbu protiv ove pandemije.
00:30
Our work was originally designed
6
30078
1863
Naš rad je prvotno bio namijenjen
00:31
to tackle the antibiotic resistance crisis.
7
31965
2853
rješavanju krize uslijed otpornosti na antibiotike.
00:34
Our project seeks to harness the power of machine learning
8
34842
4689
Cilj našeg projekta je iskoristiti moć strojnog učenja
00:39
to replenish our antibiotic arsenal
9
39555
1846
da se dopuni arsenal antibiotika
00:41
and avoid a globally devastating postantibiotic era.
10
41425
3838
i izbjegne globalno pogubno postantibiotsko doba.
00:45
Importantly, the same technology can be used
11
45685
2820
A što je važno, ta ista tehnologija se može koristiti
00:48
to search for antiviral compounds
12
48529
2072
za traženje antivirusnih spojeva
00:50
that could help us fight the current pandemic.
13
50625
2678
koji bi nam mogli pomoći u borbi protiv aktualne pandemije.
00:54
Machine learning is turning the traditional model of drug discovery
14
54080
3902
Strojno učenje izvrće dosadašnji model otkrivanja lijekova
00:58
on its head.
15
58006
1404
naglavačke.
00:59
With this approach,
16
59434
1225
Ovakvim pristupom,
01:00
instead of painstakingly testing thousands of existing molecules
17
60683
4078
umjesto mukotrpnog laboratorijskog testiranja tisuća postojećih molekula,
01:04
one by one in a lab
18
64785
1436
jedne po jedne
01:06
for their effectiveness,
19
66245
1587
na njihovu učinkovitost,
01:07
we can train a computer to explore the exponentially larger space
20
67856
4657
možemo obučiti računalo da istraži eksponencijalno veći prostor
01:12
of essentially all possible molecules that could be synthesized,
21
72537
3584
praktično svih mogućih molekula koje bi se mogle sintetizirati,
01:16
and thus, instead of looking for a needle in a haystack,
22
76145
5614
i tako, umjesto da tražimo iglu u stogu sijena,
01:21
we can use the giant magnet of computing power
23
81783
3760
možemo iskoristiti moć računala da kao ogroman magnet
01:25
to find many needles in multiple haystacks simultaneously.
24
85567
3915
privuče mnogo igala iz više stogova odjednom.
01:30
We've already had some early success.
25
90423
1992
Već smo ostvarili neki početni uspjeh.
01:33
Recently, we used machine learning to discover new antibiotics
26
93010
5465
Nedavno smo koristili strojno učenje da bismo otkrili nove antibiotike
01:38
that can help us fight off the bacterial infections
27
98499
2560
koji nam mogu pomoći u borbi protiv bakterijskih infekcija
01:41
that can occur alongside SARS-CoV-2 infections.
28
101083
3611
koje se mogu pojaviti uz SARS-CoV-2 infekcije.
01:45
Two months ago, TED's Audacious Project approved funding for us
29
105181
4169
Prije dva mjeseca, TED-ov Audacious Project nam je odobrio sredstva
01:49
to massively scale up our work
30
109374
2188
kako bismo znatno povećali svoj rad
01:51
with the goal of discovering seven new classes of antibiotics
31
111586
4628
s ciljem pronalaska sedam novih klasa antibiotika
01:56
against seven of the world's deadly bacterial pathogens
32
116238
3483
protiv sedam najsmrtonosnijih bakterijskih patogena na svijetu
01:59
over the next seven years.
33
119745
2055
u idućih sedam godina.
02:02
For context:
34
122206
1733
Radi pojašnjenja:
02:03
the number of new class of antibiotics
35
123963
1928
broj novih klasa antibiotika
02:05
that have been discovered over the last three decades is zero.
36
125915
3235
otkrivenih u posljednja tri desetljeća je nula.
02:10
While the quest for new antibiotics is for our medium-term future,
37
130030
3571
Dok je potraga za novim antibioticima srednjoročni zadatak za našu budućnost,
02:13
the novel coronavirus poses an immediate deadly threat,
38
133625
4652
novi koronavirus predstavlja neposrednu smrtnu prijetnju,
02:18
and I'm excited to share that we think we can use the same technology
39
138301
3793
i drago mi je reći da mislimo da možemo koristiti istu tehnologiju
02:22
to search for therapeutics to fight this virus.
40
142118
2809
u potrazi za terapijama za obranu od ovog virusa.
02:25
So how are we going to do it?
41
145486
1719
Pa, kako ćemo to učiniti?
02:27
Well, we're creating a compound training library
42
147229
2948
Dakle, mi pravimo jednu objedinjenu knjižnicu uzoraka
02:30
and with collaborators applying these molecules to SARS-CoV-2-infected cells
43
150201
5542
i sa suradnicima primjenjujemo ove molekule na stanice inficirane SARS-CoV-2
02:35
to see which of them exhibit effective activity.
44
155767
3894
da vidimo koje su od njih učinkovite.
02:40
These data will be use to train a machine learning model
45
160175
3192
Ovi podaci će se koristiti za osposobljavanje modela strojnog učenja
02:43
that will be applied to an in silico library of over a billion molecules
46
163391
4070
koji će se unijeti u 'in silico' knjižnicu od preko milijardu molekula,
02:47
to search for potential novel antiviral compounds.
47
167485
4204
u potrazi za potencijalnim novim antivirusnim spojevima.
02:52
We will synthesize and test the top predictions
48
172324
2658
Sintetizirat ćemo i testirati najbolja predviđanja
02:55
and advance the most promising candidates into the clinic.
49
175006
2889
i na najperspektivnijim kandidatima sprovesti klinička testiranja.
02:58
Sound too good to be true?
50
178356
1778
Zvuči isuviše dobro da bi bilo istinito?
03:00
Well, it shouldn't.
51
180158
1432
Pa, ne bi trebalo.
03:01
The Antibiotics AI Project is founded on our proof of concept research
52
181614
3325
Projekt Antibiotici pomoću UI zasniva se na dokazu koncepta istraživanja
03:04
that led to the discovery of a novel broad-spectrum antibiotic
53
184963
3401
koji je doveo do otkrića novog antibiotika širokog spektra
03:08
called halicin.
54
188388
1185
po imenu Halocin.
03:10
Halicin has potent antibacterial activity
55
190443
2813
Halocin ima vrlo jako antibakterijsko djelovanje
03:13
against almost all antibiotic-resistant bacterial pathogens,
56
193280
4102
na skoro sve bakterijske patogene otporne na antibiotike,
03:17
including untreatable panresistant infections.
57
197406
3641
uključujući neizlječive panrezistentne infekcije.
03:21
Importantly, in contrast to current antibiotics,
58
201862
2270
Što je bitno, za razliku od sadašnjih antibiotika,
03:24
the frequency at which bacteria develop resistance against halicin
59
204156
3694
učestalost kojom bakterije razvijaju otpornost na Halocin
03:27
is remarkably low.
60
207874
1484
je iznimno niska.
03:30
We tested the ability of bacteria to evolve resistance against halicin
61
210303
4710
Testirali smo sposobnost bakterija da razviju otpornost na Halocin,
03:35
as well as Cipro in the lab.
62
215037
1788
kao i na Cipro, u laboratoriju.
03:37
In the case of Cipro,
63
217299
1542
Kad je u pitanju Cipro,
03:38
after just one day, we saw resistance.
64
218865
2825
uočili smo otpornost poslije samo jednog dana.
03:42
In the case of halicin,
65
222213
1478
U slučaju Halocina,
03:43
after one day, we didn't see any resistance.
66
223715
2115
nakon jednog dana nismo vidjeli nikakvu rezistentnost.
03:46
Amazingly, after even 30 days,
67
226479
3302
Začudo, nakon čak 30 dana,
03:49
we didn't see any resistance against halicin.
68
229805
2601
i dalje nismo uočili nikakvu otpornost na Halocin.
03:53
In this pilot project, we first tested roughly 2,500 compounds against E. coli.
69
233098
5526
U ovom pilot projektu smo najprije okvirno testirali 2 500 spojeva na Е. coli.
03:59
This training set included known antibiotics,
70
239259
2780
U probni postupak su bili uključeni poznati antibiotici,
04:02
such as Cipro and penicillin,
71
242063
1746
kao što su Cipro i penicilin,
04:03
as well as many drugs that are not antibiotics.
72
243833
2272
kao i mnogi drugi lijekovi koji nisu antibiotici.
04:06
These data we used to train a model
73
246984
2587
Ove podatke smo koristili da bismo osposobili model
04:09
to learn molecular features associated with antibacterial activity.
74
249595
3978
da nauči molekularne osobine vezane za antibakterijsku aktivnost.
04:14
We then applied this model to a drug-repurposing library
75
254269
2701
Onda smo taj model unijeli u knjižnicu lijekova za prenamjenu
04:16
consisting of several thousand molecules
76
256994
2478
koja se sastoji od nekoliko tisuća molekula
04:19
and asked the model to identify molecules
77
259496
2618
i zadali modelu da identificira molekule
04:22
that are predicted to have antibacterial properties
78
262138
2784
za koje se predviđa da imaju antibakterijska svojstva,
04:24
but don't look like existing antibiotics.
79
264946
2473
ali ne liče na postojeće antibiotike.
04:28
Interestingly, only one molecule in that library fit these criteria,
80
268427
4797
Interesantno, samo jedna molekula u toj knjižnici odgovara ovim kriterijima
04:33
and that molecule turned out to be halicin.
81
273248
2336
i ispostavilo se da je ta molekula Halocin.
04:36
Given that halicin does not look like any existing antibiotic,
82
276444
3088
Obzirom da Halocin ne liči ni na jedan drugi postojeći antibiotik,
04:39
it would have been impossible for a human, including an antibiotic expert,
83
279556
4154
bilo bi nemoguće da čovjek, pri tome misleći i na stručnjaka za antibiotike,
04:43
to identify halicin in this manner.
84
283734
2184
identificira Halocin na ovaj način.
04:46
Imagine now what we could do with this technology
85
286574
2630
Sada zamislite što bismo mogli činiti ovom tehnologijom
04:49
against SARS-CoV-2.
86
289228
1741
u borbi protiv SARS-CoV-2.
04:51
And that's not all.
87
291783
1365
I to nije sve.
04:53
We're also using the tools of synthetic biology,
88
293172
2820
Također koristimo alate za sintetičku biologiju,
04:56
tinkering with DNA and other cellular machinery,
89
296016
2611
eksperimentirajući s DNK i drugom staničnom mašinerijom,
04:58
to serve human purposes like combating COVID-19,
90
298651
3910
sve s ciljem pomoći ljudima kao što je borba protiv COVID-19
05:02
and of note, we are working to develop a protective mask
91
302585
3647
i, kao napomena, radimo na razvoju zaštitne maske
05:06
that can also serve as a rapid diagnostic test.
92
306256
3432
koja također može služiti kao brzo dijagnostičko sredstvo.
05:10
So how does that work?
93
310192
1472
А kako ona funkcionira?
05:11
Well, we recently showed
94
311688
1205
Pa, nedavno smo pokazali
05:12
that you can take the cellular machinery out of a living cell
95
312917
2943
da se stanična mašinerija može izvući iz žive stanice
05:15
and freeze-dry it along with RNA sensors onto paper
96
315884
4092
i izvršiti liofilizacija zajedno s RNK senzorima na papir
05:20
in order to create low-cost diagnostics for Ebola and Zika.
97
320000
4916
da bi se dobila jeftina dijagnostika za ebolu i zika virus.
05:25
The sensors are activated when they're rehydrated by a patient sample
98
325503
5227
Senzori se aktiviraju kada ih rehidrira uzorak pacijenta
05:30
that could consist of blood or saliva, for example.
99
330754
2822
koji se može sastojati, na primjer, od krvi ili pljuvačke.
05:33
It turns out, this technology is not limited to paper
100
333600
3261
Ispostavilo se da ova tehnologija nije ograničena samo na papir,
05:36
and can be applied to other materials, including cloth.
101
336885
2886
nego se može primijeniti i na druge materijale, uključujući platno.
05:40
For the COVID-19 pandemic,
102
340671
1942
Za pandemiju COVID-19
05:42
we're designing RNA sensors to detect the virus
103
342637
4346
dizajniramo RNK senzore da otkriju virus
05:47
and freeze-drying these along with the needed cellular machinery
104
347007
3210
i liofiliziramo ih zajedno s potrebnom staničnom mašinerijom
05:50
into the fabric of a face mask,
105
350241
2707
u platno maske za lice,
05:52
where the simple act of breathing,
106
352972
2229
gdje jednostavni čin disanja,
05:55
along with the water vapor that comes with it,
107
355225
2277
uz vodenu paru koja se pri njemu podrazumijeva,
05:57
can activate the test.
108
357526
1760
može aktivirati test.
05:59
Thus, if a patient is infected with SARS-CoV-2,
109
359804
4260
Tako, ako je pacijent inficiran SARS-CoV-2,
06:04
the mask will produce a fluorescent signal
110
364088
2073
maska će proizvesti fluorescentni signal
06:06
that could be detected by a simple, inexpensive handheld device.
111
366185
3830
koji se može detektirati običnim jeftinim ručnim uređajem.
06:10
In one or two hours, a patient could thus be diagnosed
112
370534
4484
Za dva ili tri sata pacijentu bi se tako mogla postaviti dijagnoza
06:15
safely, remotely and accurately.
113
375042
2972
na siguran i točan način, bez kontakta.
06:18
We're also using synthetic biology
114
378735
2520
Тakođer koristimo sintetičku biologiju
06:21
to design a candidate vaccine for COVID-19.
115
381279
2720
da osmislimo cjepivo protiv COVID-19.
06:25
We are repurposing the BCG vaccine,
116
385014
2653
Vršimo prenamjenu BCG cjepiva
06:27
which had been used against TB for almost a century.
117
387691
2870
koje se koristi protiv TBC skoro čitavo stoljeće.
06:30
It's a live attenuated vaccine,
118
390585
1541
То је živo oslabljeno cjepivo
06:32
and we're engineering it to express SARS-CoV-2 antigens,
119
392150
4657
i dizajniramo ga tako da sadrži SARS-CoV-2 antigene,
06:36
which should trigger the production of protective antibodies
120
396831
2814
koji bi trebali potaknuti imunosni sustav
06:39
by the immune system.
121
399669
1635
da proizvodi zaštitna antitijela.
06:41
Importantly, BCG is massively scalable
122
401328
2734
A što je važno, BCG je izrazito skalabilan
06:44
and has a safety profile that's among the best of any reported vaccine.
123
404086
4573
i ima sigurnosni profil koji je među najboljim od svih registriranih cjepiva.
06:49
With the tools of synthetic biology and artificial intelligence,
124
409881
5105
Pomoću alata sintetičke biologije i umjetne inteligencije,
06:55
we can win the fight against this novel coronavirus.
125
415010
3348
možemo pobijediti u borbi protiv ovog novog koronavirusa.
06:58
This work is in its very early stages, but the promise is real.
126
418844
3319
Ovaj rad je u svojoj veoma ranoj fazi, ali očekivanje je realno.
07:02
Science and technology can give us an important advantage
127
422798
3445
Znanost i tehnologija nam mogu dati jednu važnu prednost
07:06
in the battle of human wits versus the genes of superbugs,
128
426267
3161
u borbi između ljudske pameti i gena rezistentnih bakterija,
07:09
a battle we can win.
129
429452
1747
borbi u kojoj možemo pobijediti.
07:11
Thank you.
130
431990
1233
Hvala.
O ovoj web stranici

Ova stranica će vas upoznati s YouTube videozapisima koji su korisni za učenje engleskog jezika. Vidjet ćete lekcije engleskog koje vode vrhunski profesori iz cijelog svijeta. Dvaput kliknite na engleske titlove prikazane na svakoj video stranici da biste reproducirali video s tog mjesta. Titlovi se pomiču sinkronizirano s reprodukcijom videozapisa. Ako imate bilo kakvih komentara ili zahtjeva, obratite nam se putem ovog obrasca za kontakt.

https://forms.gle/WvT1wiN1qDtmnspy7