start	end	text
0	2260	Okay, we're live. Take it away.
3800	5840	Hi. Hi there.
6620	7760	Can they see us?
7760	10480	Yes, I guess so. Let's see whether they can...
10480	11920	You are live. You're live.
11920	14460	We are live. Okay, so here we are.
14460	16460	Hi everyone. Hi. I'm Noa.
16460	17820	This is Lawrence.
17860	19600	Nice to meet you. We are here
20180	22380	streaming from the Weissman Institute of Science
22380	23800	in Israel. And
23800	25800	together we are going to present
26700	27180	the
28500	29460	comparison between
29460	31160	cybersecurity and biology.
31160	32340	I hope you are ready.
32800	35180	Well, I'm ready, so I guess the stage is yours, Noa.
35180	35980	Okay.
37940	39260	So, nice to
39260	40620	meet you. I will just
42720	43200	present...
43200	45360	Okay. Sorry.
45720	47100	Okay, so our agenda
47100	49080	for today... I have two
49080	51020	targets with my presentation.
51020	53180	First of all, I'll show you how beautiful and complex
53180	55100	biology is. And the second
55100	57300	part will be to make you jealous.
57600	59120	So, who am I?
59120	61080	My name is Noa. I'm a master
61080	63320	student at the Weissman Institute of Science.
63320	64720	I'm a junior scientist in
65780	66380	a lab
67240	68820	that is doing research
68820	71820	specifically in microbiome, but also diabetes.
72040	73080	And I'm a former
73080	75600	cybersecurity researcher at
75600	76800	Checkpoint.
78480	79380	So,
79620	81880	until a year ago,
81880	83380	I was a
83380	85300	reverser in Checkpoint. My job
85300	86380	was to reverse...
87500	90320	I'm sure that you all know about this job,
90320	91600	to reverse engineering malware,
91900	93320	and then to implement
94140	95260	the defensive
95260	97300	coding into Checkpoint
98240	99360	antivirus.
99960	101260	So, you know, I thought
101260	104100	I'm a pretty smart girl, after all.
104160	105260	I even did
105260	107340	some fuzzing stuff. I found some
107340	108300	vulnerabilities.
110260	111340	You know, I
111340	113780	built myself up as a
113780	115640	hacker, as a researcher
115640	117860	in Checkpoint for many years.
121040	122440	And then...
122440	123700	But I was always
123700	125820	very curious about complex
125820	128020	themes, not only malwares, but also
128020	130300	about biology.
130460	131760	And a year ago, I decided
131760	134940	to come here and to become
134940	137260	a researcher in biology,
137260	138660	specifically about
138660	139320	the new area in
139320	140760	microbiome.
140860	143240	So I thought, you know, I'm so cool,
143240	144500	and I'm doing reverse engineering
145440	147400	for so many years.
147460	149120	So, you know, all of the
149120	151400	bacteria and the COVID-19 and whatever
151940	153120	comes up,
153120	154700	all of them
154700	156040	are going to be
157020	158960	scared of me, because now
158960	161040	there is a good reverser and a good hacker
161040	162780	that is coming into the field.
162880	164720	But then I found out that
166000	167440	apparently it's not as...
167440	169540	I think that we all...
169540	171540	I think that it will be really
171540	173520	easy for me to prove my point that it's
173520	175440	not as easy as we can...
176080	177440	as we thought,
177440	180120	until not very far.
180700	181840	You can
181840	183540	see, like, what happened when
183540	185360	just one bacteria just
185360	187300	did some
188100	189300	evolution and become
189300	190660	just a little bit
191380	193560	more dangerous than before.
193560	195520	Like, the COVID
195520	197560	is doing evolution
197560	199520	for many years and
199520	201780	like, I think it's really
201780	203900	easy to see how does it affect
203900	205660	the world right now.
206720	207900	And also
207900	209940	how I learned in a hard way
209940	211820	biology is way more
211820	213840	complex than any malware that I ever did
213980	215840	a reverse engineering in. This is
215840	217740	actually one...
217740	219840	just one slide of my
219840	221980	class. What you see here is
221980	224000	the complex system
224000	225880	of what happens inside
226100	227720	a cell when there is one
227720	229740	ligand that is binding
229740	231360	one of the receptors.
232420	233700	So, apparently
233700	236300	it wasn't the best idea for me.
236560	238300	I'm joking, of course.
238360	239940	I'm very curious and
239940	241720	this area is very challenging
241720	243320	for me.
243760	245960	And I think
245960	247640	that what we do here is just
247640	248680	beautiful.
250860	251880	And also
251880	253760	one more thing that I want to
253760	255740	point out during my lecture is that
255740	257660	when I tell people about biology
257660	259060	they all think that I'm
259840	261100	palpating and growing
262160	263740	microbes on a plate
263740	265020	but actually these days
266120	267540	computational biology is
267540	269040	on the rise and what we are doing
269680	272040	especially in Weizmann
272040	273740	is trying to take
273740	275800	each of the biology systems that we have
275800	277760	in our body or in bacteria or anything else
277760	279580	and build a model about it. This is
279580	281700	also one of my classes and this is from
281700	283720	Uri Alonlav that we are
283720	285420	doing a collaboration with.
285420	288200	So what he did here, he took
288200	289200	the aldenal gland
289200	291520	and tried to build
291520	293080	actual equations
293080	295440	I think I did something
295440	297700	okay, and tried to build
297700	299320	equations in order to
299320	300740	understand that this
301480	302900	thing here is the
304180	305620	complex system of
305620	307580	the glands and this one
307580	309760	is some
309760	311580	equations that are
311580	313520	explaining why
313520	315180	does diabetes type 2
316080	316600	develop
317760	320020	in so many people these days.
320800	321680	And now I will pass
321680	323400	the microphone to my friend that will
323400	324900	explain to you in detail about
325400	327540	some of the basic creatures that all of us
327540	328720	have in our body.
329880	332140	Thank you very much indeed Noor.
332760	333760	By the way, I just
333760	335180	want to point out that
335180	336860	the person
337320	339060	that actually introduced me
339060	341060	into this world and made me
341060	342660	very stressed
344600	345400	also a very
345400	347300	happy student but also a very stressed and a very
347300	349420	poor student is this guy.
349420	351220	So when your mom tells you not to talk
351220	353360	to strangers, maybe she has a point.
354440	355280	Okay, so I
355280	357040	guess I don't need any further ado.
357040	359080	Well, if you can see the camera, you will see that
359080	361320	in contrast to Noor, I will be sitting. I'm a little older
361320	363780	so forgive me for that.
364680	366040	So actually
366040	368080	when we were preparing this lecture
369120	369800	we
370560	371240	we
374520	376060	we came up
376060	378180	first with this title
378180	379680	Back to the Future
379680	381660	and me being this
381660	384080	very modest, humble
384080	386060	German dude that I am. I didn't
386060	388000	have all these fancy animations
388000	390020	that Noor showed to you.
390020	392000	But actually when I envisioned this presentation
392000	393700	I started with something
393700	395720	lame looking like this.
395720	397980	So again,
397980	399620	we are from the Advisory Institute of Science
399620	401860	in Israel and
403080	403760	what
403760	405940	I want to show you now
405940	408280	in the next 25 minutes
408280	409760	is in essence
409760	411880	how computer science
411880	413860	and biology not only go
413860	416240	side by side exhibiting stunning parallels
416240	417860	but also that they in fact
417860	419860	have an intersection that we can
419860	422020	exploit nowadays to address
422020	423960	current challenges that we
423960	425380	are facing.
425460	427540	But before I go into this
427540	429960	I actually want to fulfill
429960	431520	another request that Noor
431520	433960	kindly asked me which is
433960	435720	to explain a very central
435720	437880	concept in biology which by
437880	439420	itself also has
441540	441940	a counterpart
441940	443260	in computer science
443690	446800	and I'm talking about genetic code.
446800	447620	Okay?
447620	449260	Genetic code, our DNA
449260	451660	is basically the language
451660	453760	in which life is written.
453760	455100	So that's sort of
455420	457720	these chemical molecules that make up
457720	458840	four letters
459500	461300	A, T, C and G.
461300	463840	They write a
464320	465660	blueprint of life
466920	467480	but as such
467480	469840	they are not very useful because they have to be
469840	472040	transcribed. They have to be transcribed
472040	474260	from this DNA form, this genetic
474260	475100	information into just another
475100	476860	sort of code
476860	479300	which is RNA which then
479300	481580	is sort of translated
481580	482700	so put into
483660	485040	proteins, protein
485040	487200	molecules that make up all our cells
487200	488560	and therefore our body
489100	491040	as a central component.
491420	493290	And speaking about code, I think
493740	495280	again, it's not very far
495280	496960	fetched to reach from
496960	499040	biology, so the science
499040	500440	of life
501100	503140	to computer science because
503140	505020	you as computer scientists, mostly
505020	506760	out there in the audience, you will know what
507360	508320	a good computer
509120	510460	code is.
511860	513000	And basically I want to start
513000	515000	by asking the question whether
515000	517320	there is another analogy between
517320	518880	computer science and
518880	521100	biology. Are there any
521100	523120	parallels in terms of computer
523120	525300	viruses and biological
525300	527220	viruses because, I mean,
527220	529280	as you know, currently we are facing
529500	529920	a global
530680	531880	COVID-19
531880	533740	pandemic that originates from the
533740	535940	deadly spread of the
535940	537780	coronavirus, SARS-CoV-2
538560	539840	and I
539840	541980	want to raise the question whether
541980	544000	there's anything, if you
544000	546080	only know computer viruses because you're
546460	547980	a cyber expert, that you can
547980	550060	also learn about biological
550060	552060	viruses such as this
552060	553460	coronavirus.
553740	555940	So let's start with a computer virus, right?
555940	557920	With a computer virus that you probably heard about
557920	560080	the Melissa back in 1999
560080	560560	which
562760	563480	based on
564180	565900	a sequence like a stretch of code that you
565900	567120	can see on the left here
568740	569960	which basically with
569960	572020	some letters just gives
572020	574300	also the blueprint of a
574300	576160	malware basically that causes
576160	578080	a lot of damage and
578080	580100	I'm not sure whether you're surprised or not but what you're
580100	582120	going to see on the right now here
582120	586080	are the first 3,569
583920	585920	letters
586080	587900	of the SARS-CoV-2
587900	590220	this coronavirus, the version
590220	592020	Wuhan-U1
592020	593960	the genome, the genetic
593960	596100	information of this
596100	597820	virus which in total is about
597820	599420	30,000 letters long
599880	601960	that makes up all
601960	604040	this virus that now
604040	605940	spreads on the face of the Earth
605940	607960	and I'm
607960	610140	saying genetic information here but to be
610140	611500	precise this is not
612080	614100	DNA that you're seeing but
614100	615680	actually it's an RNA virus
615680	618100	so while this is
618100	620040	sort of the genetic information of
620040	621440	this virus, it's not
622320	624160	the same genetic information
624160	626160	that we have in our body sense but it's
626160	627640	only this transcript that I
627640	629700	introduced to you before.
630080	631800	But looking at this
631800	633960	sequence or this source code
633960	636000	of the virus, you wouldn't be
636000	637920	able to judge that this is particularly harmful.
637920	639860	Maybe on the left you could,
639860	642000	Melissa, because you're
642000	644000	an expert and you see what
644000	644960	it actually does
646420	648000	but there is no chance
648000	649780	even the best molecular
649780	651840	biologist looking at
651840	653920	the code on the right would be able to
653920	655960	tell, okay, this is a deadly virus.
656740	657780	But still, we
657780	659860	desperately needed to know this sequence
659860	661800	to detect this virus
662720	663860	but then the question
663860	665980	is always having this sequence, having
665980	668000	this source code, what does it do?
668000	670000	What is this sequence translated
670000	672100	to? And starting again
672100	673120	with the example
674000	676020	of Melissa,
676020	677900	basically what this virus did
677900	679700	is it prompted
679700	681320	always to send
682000	683680	emails to your own
683680	685940	Microsoft Outlook inbox or
685940	687200	started new
688000	690000	Word files which basically then
690800	691600	overloaded
691600	693260	and completely flooded
694000	695940	servers, like online servers and
695940	697520	stuff. So you had,
698020	699560	of course, I only made this up like
699560	700920	an infinite inbox
703100	703540	and
703540	705320	this is sort of how
705320	707800	the source code of the computer virus
707800	709040	translates into
709900	711460	a 3D structure, if you want to say so,
711460	713960	with a malfunction, so with a bad function.
714540	715520	And actually the same thing
715520	717520	biologists did then, they asked the question
717520	719760	okay, given the sequence of the
719760	721580	coronavirus, how does this
721580	723540	translate into what I mentioned before
723540	725520	these proteins, right, that make
725520	727520	up the shell of the
727520	728560	virus and
729440	731260	these different parts of the virus that
731260	733340	help the virus to infect
733340	735680	our body. And this is
735680	737540	basically one part of this
737540	739360	virus, so it's not the entire virus, it's just
739360	741500	one tool, actually what I said
741500	743300	like one element, one
743300	745520	part of a molecule that helps this virus
745520	747460	to infect our body, our
747460	748640	body cells.
749380	751400	And this is how this virus does
751400	753460	harm, right, it infects cells and then
753460	755200	eventually kills them.
755900	757320	So this is how in biology
757640	759100	a sequence or a source code
759100	761520	translates to a structure
761520	763520	and then a certain function.
764420	765420	And then to
765420	767440	sort of visualize how it looks
767440	769280	like in total, apart from
769280	771400	what it does, the
771400	773640	Melissa virus to overload
773640	775240	your inbox and the coronavirus
775240	778760	to infect your body cells,
778760	779460	actually
779460	781360	for the Melissa virus then it took
781360	783480	an artist to visualize
783480	785340	how this overload looks like
785340	787400	and this is how an artist apparently pictures
787400	789160	how a computer virus
789160	791440	looks like, which as you
791440	793340	can probably better judge than I do
794180	795220	is just
795220	797260	more or less arbitrary
797260	798920	representation.
799440	801400	But in fact, in biology, again, we can
801400	803360	use this wonderful
803360	805340	invention of microscopes to
805340	807500	actually look at how the virus looks
807500	809660	like. However, conventional
809660	811620	light microscopes are not sufficient enough
811620	813760	here, so actually we need electron
813760	815740	microscopes. However, if we use
815740	817400	these very sophisticated
818020	819680	apparatuses that cost like
819840	821600	a million dollars each and
821600	823780	require a lot of expertise to use them,
823780	825980	then we can actually see a virus inside
825980	827880	of cells.
827880	829440	So all these blue particles
829440	831880	that in this Forbes article
831880	833100	are described as
833840	835900	pepperoni, sorry,
835900	838020	slices of sausage
838020	840000	on your pizza, they are
840000	842160	the actual virus. However, I also must
842160	843540	say the truth is
843540	846140	if you use an electron microscope to observe
846140	847000	the coronavirus
848840	850500	in cells,
850500	851960	so this is actually the coronavirus of
851960	853920	the first identified
853920	855120	US patient
855120	858240	that was imaged here,
858240	860020	this original picture from
860020	862480	the electron microscope had no color.
862720	864060	So this blue
864060	865620	color is just put up
865620	867860	artificially in the end to visualize
867860	869700	the virus particles.
870980	871720	But here
871720	873000	we are still talking about
874240	874760	only
876560	877620	the size of
878300	879720	a meter divided by
879720	881600	one billion. So it's
881600	883400	in this order of magnitude.
883400	885080	So it's very small.
886300	887540	And you need sophisticated,
887540	889420	as I said, microscopes to actually see it,
889420	891440	but you can see it. You can understand
891440	893460	how this sequence source
893460	895500	code information that we saw,
895500	897140	how this translates to
897140	899140	first individual
899140	901340	aspects of this virus,
901340	904020	individual molecules, and then the virus as such.
904020	905580	Now, once it
905580	907620	has done its terrible job infecting
907620	909300	our body cells.
909640	911580	And this sort of
912400	913480	shows that
913480	915200	to some extent there are
915200	917360	parallels between computer science and
917360	919320	biology, in this case computer
919320	920820	viruses and
920820	922780	biological viruses such as
922780	924540	SARS-CoV-2.
926200	926540	But
926540	928640	is there, you know, besides
929980	930540	from
932040	933400	this example,
933400	934960	is there something we can
934960	937040	do about this? Is there something we can use
937040	938640	from computer science
938640	940860	to help us
940860	942360	fighting this
942360	944920	terrible pandemic that we are facing
944920	947120	in the US and all over the globe?
947220	948880	Is there an intersection between computer
948880	950580	science and biological aspects of
950580	952320	this pandemic?
952420	954760	This is a question Noah and I basically
954760	956960	asked ourselves. And in the very
956960	958220	beginning of this pandemic
958700	960740	in March, in fact, we
960740	962860	participated in
963700	964840	a hackathon that was
964840	966780	initiated by
966780	968780	InDev in collaboration
968780	970560	with Microsoft. They called it
970560	972000	HackCorona.
972440	975180	Creativity will not be quarantined.
975180	976160	And in the next
976160	978520	five to ten minutes, I would like
978520	980540	to introduce to you this
980540	982500	project that we initiated there to show
982500	984620	you how computer science can actually facilitate
984620	986280	in a very
986280	988420	concrete example, our
989280	990000	fight
990000	992440	against this virus.
993340	994080	And the
995400	996420	question that
996420	998240	we asked basically is
999800	1000520	how
1000520	1002400	can we use computers to
1002400	1004280	combat COVID-19?
1004280	1006220	And the particular question we were wondering
1006220	1008280	about, because this was of major
1008280	1010420	concern in the beginning of this pandemic,
1010420	1012080	is how to
1012080	1014180	help people who develop severe
1014180	1016060	complications. Because we
1016060	1018800	understood that asymptomatic patients,
1018800	1020320	of course, they are very dangerous in
1020320	1022180	terms of unobserved
1022180	1024580	spreaders of this disease.
1024820	1026380	But in case
1026380	1028340	someone develops severe complications,
1028340	1031120	needs ventilation in the hospital,
1031120	1032380	then this
1032380	1033040	might
1034820	1036520	present a major challenge for
1036520	1038320	the health system in Israel
1038320	1040040	and all over the world.
1040680	1042560	So how can we basically assess
1042560	1044700	the uncertainty whether a certain
1044700	1046480	COVID-19 patient will develop
1046480	1047340	severe
1048520	1049820	symptoms and require
1051160	1052650	hospitalization and ventilation?
1053820	1054500	This is
1054500	1056380	the question we asked. And as you can see
1056380	1058280	in this picture, again, we were
1058280	1060440	pretty much using
1060440	1062540	biology as well as computer science
1062540	1064280	by these two mice that you can see here,
1064280	1066400	like a computer mouse and an actual mouse,
1066400	1068860	or at least a stuffed animal of a mouse.
1070540	1072320	And so, in simple
1072320	1074740	terms, our question was,
1074740	1076260	given the fact that we have
1076420	1078040	a certain number of
1078040	1079580	COVID-19 patients
1080640	1082100	that require, because
1082100	1084580	they develop severe complications,
1084580	1086060	hospital beds,
1086060	1088140	how to deal with them if the number of
1088140	1090100	hospital beds is actually
1090100	1092120	lower than the number of patients who require
1092120	1094240	hospitalization? In other
1094240	1096600	words, five
1096600	1098220	exceeds three, right?
1098220	1099780	So how to decide
1099780	1101920	which of them to allocate
1101920	1103880	to the hospital?
1103880	1105980	How to estimate what's the fraction of
1105980	1108100	people who develop severe complications
1108100	1110120	and how to decide whom to offer
1110120	1112000	with a certain priority these
1112000	1113880	available resources?
1113880	1116060	This was the question.
1116060	1117560	And the reason why we deal with this
1117560	1120140	was because at the Weizmann Institute
1120140	1121880	we are really at the forefront
1121880	1123880	of this, not only that we have
1123880	1125880	medical doctors on campus
1125880	1127880	who see these patients on a day-to-day
1127880	1129920	basis, but also the labs
1129920	1131580	we're working in from
1131580	1134080	Prof. Dr. Dr. Eran
1134080	1136360	Elinav and Prof. Dr. Ido Amit,
1136360	1137780	they develop now
1137780	1140080	technologies to not only
1142380	1143500	detect these
1143500	1145340	infections and stratify
1145340	1147460	these patients, but also streamline this
1147460	1149120	process, really linking
1149120	1151280	the foundational research with
1151280	1153760	the application in the clinics.
1154100	1154780	Okay?
1154780	1157640	So this is why we also felt
1157640	1159540	dedicated to address this
1159540	1161620	with our expertise, which
1161620	1162960	comes more from the
1163500	1165240	computational side here.
1165240	1167540	So what we did in the beginning
1167540	1169500	is, first of all, we asked, okay,
1169500	1171780	what's the data we are dealing with?
1171780	1173720	Because only if you have data,
1173720	1175800	you can actually use computational routines
1175800	1177980	to address
1177980	1179980	all these issues.
1180640	1181660	So we
1181660	1183680	basically simulated
1183680	1185920	based on existing
1185920	1187780	data,
1187780	1188640	an Israeli
1189780	1191840	dataset comprised of 10,000
1191840	1193980	cases, where we really used
1193980	1195100	Israeli demographic
1195780	1197860	information and health record data
1197860	1199400	that was publicly available
1199400	1201760	to a certain extent,
1201760	1203660	and set up
1203660	1205420	sort of a database where
1205420	1207240	we can get an idea, okay, this is Israeli
1207780	1209240	population, and now
1209240	1210220	given all this
1211680	1213700	demographic information and health record information,
1213700	1214860	can we basically now
1215680	1217360	estimate what's the fraction of people who
1217360	1219620	might develop severe COVID-19
1219620	1221180	complications? So first of all,
1221180	1223360	as you can see, we are struggling entering all this
1223360	1225280	data, but then what we obtained
1225280	1227100	is basically a more or less realistic
1227100	1229160	Israeli age distribution
1229160	1231340	here, and
1231340	1233160	also what seems to be
1233160	1234840	important, and we will hear about
1234840	1237180	diabetes and the metabolic syndrome later
1237180	1238740	on in Noah's part,
1238740	1241340	the distribution of the body mass
1241340	1242960	index in the Israeli
1242960	1244580	population,
1245100	1247420	which is basically a measure for your
1247420	1249360	body density,
1249360	1251260	so it's your body size divided
1251260	1252340	by your
1253080	1255280	body height to the
1255280	1257300	square, and it gives you an estimate
1258760	1259160	of...
1259640	1261360	if you have a high BMI,
1261360	1263180	you are considered obese, and then actually
1263180	1265040	this also affects your risk of
1265040	1267340	COVID-19 severe
1267340	1269080	complications, but also whether you get infected
1269080	1271260	in the first place. But it's not fully
1271260	1273060	understood, it's just an observation, so
1273060	1275300	anyways, we thought we need to include
1275300	1277400	this information, so first of all, we needed to have
1277400	1279360	some sort of realistic distribution
1279960	1281240	of this kind of
1281240	1283320	data. And then
1283320	1285520	what we were basically sampling from this information
1285520	1287100	we had back then when Israel
1287100	1289440	was not really facing many
1289440	1291240	many cases at the time
1291240	1293180	was this sort of distribution
1293180	1295240	of people
1295240	1297420	who got infected and then
1297420	1298700	sort of were
1299320	1301340	surviving through the disease, and
1301340	1303380	people who eventually died, and then you can
1303380	1305300	see that even though the overall frequency
1305300	1307460	of dead people, of course, or of dying
1307460	1309320	people is with the
1309320	1311260	infection or upon infection,
1311260	1313560	not necessarily caused by the infection,
1313560	1315340	is very low, but you can see that
1316880	1318720	if you're older, the
1319260	1321300	probability that you actually die
1321300	1323680	after getting infected with SARS-CoV-2
1323680	1325660	is higher.
1327180	1328720	So what we then
1329140	1331260	did is we used this data set,
1331260	1333060	this information that we
1333060	1335220	compiled, and
1335220	1337580	trained the model to basically
1337580	1339180	predict, given
1339520	1340640	a certain patient with
1341100	1343040	this health record and demographic
1343040	1344740	information,
1344740	1346880	what's the risk of developing
1346880	1349220	severe complications, and
1349220	1351200	then what you can always do is you can compare
1351200	1353280	your prediction to the actually observed
1353280	1355120	risk, because you have
1355120	1356660	information on this, right?
1356820	1359080	And then we get a pretty good agreement between
1359080	1360960	the predicted risk and the observed
1360960	1362340	risk here in this
1363040	1364780	preliminary data set.
1365100	1367300	And then we can ask the question, okay, what kind of information
1367300	1368860	is actually necessary to
1369700	1371620	predict this risk in a
1371620	1373460	reliable manner? And then
1373880	1375860	just as a test case,
1375860	1377640	proof of principle, we said,
1377640	1380100	okay, age seems to be
1380100	1381420	relevant and
1382820	1383680	critical,
1383680	1385320	so what happens if
1386600	1387460	we are
1387460	1389400	leaving out the age information? And what you
1389400	1391620	are seeing here is basically a receiver
1391620	1393940	operating characteristic curve,
1393940	1395160	a rock plot,
1395160	1397120	and basically if your
1397120	1399620	line, so your specificity
1399620	1400280	and sensitivity
1401240	1403440	of this prediction performs
1403440	1405220	pretty much randomly,
1405220	1407240	then this line that you see is close
1407240	1409320	to the diagonal line. So as you can see,
1409320	1412020	the orange line, if we leave out the age information,
1412020	1413220	we cannot really predict
1413220	1415200	that it's like randomly drawing,
1415200	1417240	okay, this person will develop
1417240	1419340	complications or this person
1419340	1421420	won't, but if we just consider
1421420	1423200	the age information as well in addition
1423200	1425140	to all the other information, then
1425820	1427220	this prediction becomes
1427220	1429380	very sensitive and very specific.
1429380	1431200	So it means that if we
1431200	1433120	say a person does
1433120	1434800	develop complications, then
1435360	1436980	the chance that this is a true
1436980	1438680	positive indication is
1439900	1441260	high. But also
1441260	1443620	if we say this person won't develop complications,
1443620	1445120	then also the chances
1445120	1447160	that this is a true negative, so it
1447160	1449300	indeed won't develop complications,
1449300	1451180	it's also sort of high.
1451180	1452860	So of course the perfect prediction
1452860	1455020	would be just going straight up
1455020	1456920	and then straight to the right,
1456920	1458980	but also in this toy model
1458980	1461080	or preliminary thing,
1461080	1462600	the performance of this prediction
1462600	1462980	was
1465160	1467380	not so bad, I would say.
1467660	1468700	So the way we
1468700	1470280	envision this basically is
1471140	1472760	once people get sick or
1472760	1475200	are suspected of being infected,
1475200	1476340	they have to see a doctor
1476340	1478780	who then puts in
1478780	1480700	all the information into a certain
1480700	1482780	database and some
1482780	1484060	device that collects
1484640	1486960	this information and then basically trains
1486960	1488880	this model that is based
1488880	1490320	on machine learning algorithms
1491300	1492800	and all these health record
1492800	1495020	information and then basically puts out
1495020	1496260	back to the doctor
1497520	1498880	a certain risk score.
1498880	1500620	So the application
1500620	1502300	tells the doctor, okay, this person
1502300	1504820	exhibits a certain probability to develop
1504820	1506940	severe complications, so better save a hospital
1506940	1508920	bed for this one, or in the ideal case
1510580	1512460	this person can just be
1513300	1514920	isolated at home, there is
1515080	1517600	a very low chance that it's going to be severe,
1517600	1519360	just monitor it carefully,
1519360	1520200	but most likely
1520800	1522800	the hospital bed allocated to this
1523660	1524980	patient can be
1524980	1527300	given to somebody else with a higher risk.
1528860	1530840	So, wait, what are we
1531580	1533100	lacking here? I already
1533100	1534960	presented to you that this machine learning algorithm
1534960	1536980	sort of works, but we need
1536980	1539060	this device, right? We need this application.
1539600	1540720	And this is also what we
1540720	1542520	sort of developed, like
1543080	1545100	something that can work on your mobile phone
1545100	1546920	and helps medical doctors, again, that
1546920	1548940	we have in our group and we talk to them on a daily
1548940	1551020	basis, to just use
1551020	1552940	it in a very handy way. If they
1552940	1554320	see the patient, they can check
1554740	1557200	in all this information in less than a
1557200	1559140	minute, and
1559140	1560920	then basically we can
1563220	1564220	just let
1564220	1566580	the machine learning algorithm do
1566580	1568600	its job and see whether there's
1569020	1570140	a certain risk.
1570660	1572020	And, of course, this
1572660	1574300	machine learning algorithm that I presented
1574300	1576360	to you, this model that we used there to make
1576360	1578380	the predictions, was just the first attempt
1578380	1580240	that we had in the Sakaton, you know, very
1580240	1581160	coarse-grained
1582120	1583880	first rough trial.
1585320	1586320	Needless to
1586320	1587800	say, you can also use
1587800	1590080	more sophisticated methods
1590080	1591300	where you can use all these
1592480	1592880	different
1594440	1596100	biological information on these patients
1596100	1598200	to assign how
1598200	1600140	important is this
1601760	1602160	information
1602160	1604260	to either say that there is a high
1604260	1606200	risk, or maybe it's just informative to say that there's
1606300	1608200	a low risk, or maybe a combination of both,
1608200	1610080	right? So we can, of course, you
1610080	1612180	always want more data. We can try
1612180	1614220	different models, and as
1614220	1616340	I said, we are in day-to-day contact with the clinicians
1616340	1618160	actually in charge, actually doing this, actually
1618160	1620000	doing the anamnesis or talking to the
1620000	1621800	patients,
1621800	1624500	suspected infected people.
1625720	1625860	So
1625860	1628140	this is what we envisioned,
1628140	1629960	you know, we can't really
1629960	1632220	not necessarily change
1632220	1634220	the number of people who are
1634220	1636000	susceptible and eventually will get
1636000	1638520	infected. It's just a matter of time,
1638520	1639860	it seems like.
1641160	1642200	And maybe
1642200	1643940	also, you know, the health system is sort of
1643940	1646020	currently at its limit, so we cannot
1646020	1648800	really expand the number of hospital beds.
1649020	1650120	But this app
1650120	1651620	that we developed, that by the way we
1651620	1653900	called Virologens, because it's like an
1653900	1656160	intelligent solution to virology
1656160	1657380	and inference of
1658460	1659000	this
1660800	1661880	virus-induced
1661880	1664040	health complications,
1664040	1666320	we could now use to assign
1666320	1667900	again hospital
1667900	1670060	beds to people who need it the most.
1670060	1671960	So in the scope of this hackathon, which of course is
1671960	1674320	very preliminary, and we are still sort of
1675380	1676480	thinking of how to
1676480	1678180	put this further,
1678180	1680100	in the scope of this hackathon, this
1680520	1681920	question was solved.
1682440	1684900	But as I already indicated, in the scope of this hackathon,
1684900	1686440	and this is just COVID-19, which
1686440	1688380	at the moment seems to be a major part of our
1688380	1690480	lives, but certainly it's not the only part
1690480	1692340	of our life, so
1693860	1694600	can we
1695020	1696340	make a more general
1696940	1698340	observation when we look at
1698340	1700700	computer science and biology?
1702040	1702940	Or in
1702940	1704800	other words, how can we
1705600	1706940	combine powerful
1706940	1708960	algorithms from computer science
1708960	1711200	with biological data sets?
1711200	1712860	And this is what I've shown,
1712860	1714660	what I'm showing you here, it's basically
1714660	1716680	the intersection of biological
1716680	1718780	data, this is how biological data
1718780	1721000	nowadays sort of looks like.
1722060	1722920	How can
1722920	1724880	we combine this biology with
1724880	1726740	this more like a conceptual or
1726740	1728340	theoretical view of
1728340	1730720	biology as a system?
1731080	1732480	As a living
1732480	1734620	system that just operates
1734620	1736620	certain commands and has
1736620	1738540	some input-output functions.
1738840	1740480	And the answer here is also the computer,
1740480	1742400	because only the computer will help us to process
1742400	1744300	this data and to put it into these
1744300	1746360	sort of models in terms of
1746360	1748480	understanding how biological systems
1748480	1750540	work. And this is a discipline
1750540	1752540	that I actually studied, I did my
1752540	1754720	PhD in this field,
1754720	1756880	which is called systems biology.
1757760	1758880	And let me
1758880	1760740	introduce to you the field of systems
1760740	1762960	biology with one of the godfathers of systems
1762960	1764780	biology, a guy who actually does
1764780	1766740	also a lot of robotics and
1766740	1769020	intelligent design and stuff, a Japanese
1769020	1770920	guy who also invented
1770920	1772960	robots that help people
1772960	1774720	to rescue
1774720	1776160	people from
1777140	1778320	collapsed buildings
1779080	1781160	after earthquakes and stuff.
1782360	1782960	Hiroaki
1782960	1784820	Kitano is his name, and
1784820	1786700	he describes systems biology as the
1786700	1788880	integration of experimental and
1788880	1791000	computational approaches to precisely
1791000	1792620	characterize biological systems
1792620	1795020	with quantitative data.
1795100	1796260	Okay, so now we have
1796680	1799120	quantitative data, which is machine-readable
1799120	1800220	from biology,
1800620	1803020	that we will use
1803020	1805100	to understand, in combination
1805100	1807000	with experiments in the laboratory,
1807740	1808880	how this
1808880	1810860	biological system works, simulating it at the
1810860	1812860	computer, using algorithms to
1812860	1814960	understand what's going on
1814960	1816640	there. And what does
1816640	1819040	go on there? How does this look like?
1819560	1820940	Well, you can imagine
1820940	1822840	as this sort of simple diagram
1822840	1824700	here, where basically on the
1824700	1826560	x-axis, you will have
1826560	1828880	time, and on the y-axis,
1828880	1830880	you will have the amount, quantitative
1830880	1833000	data, okay, of a biological
1833000	1835340	entity. Now,
1836780	1838660	what does this mean? So basically,
1838660	1840840	what you have in blue, these blue dots, these are
1840840	1842860	your data points, and then
1842860	1845020	with the computer, you can basically simulate the behavior
1845020	1846660	of the system, and you have some mean
1846660	1849180	simulation here, the blue line,
1849180	1850880	but then also you have this shaded region, which is
1850880	1852840	basically your uncertainty, okay,
1852840	1854860	your confidence interval. So not
1854860	1857100	only you get an understanding with the computer,
1857100	1859020	how does the system change over time,
1859020	1860760	how do biological entities
1860760	1862660	change over time,
1862660	1864900	but also, how certain are
1864900	1866680	you that this happens, right? The virus can
1866680	1868840	spread, like,
1868840	1870780	enormously, but maybe there's
1870780	1872860	also a chance it doesn't spread, or maybe there's a chance
1872860	1874980	that it spreads, like, even more.
1875240	1876840	And this is something we can quantify,
1876840	1878900	we can calculate. Sometimes
1878900	1880840	we have a problem with communicating this, but
1880840	1882840	there are possibilities in mathematics
1882840	1884880	to calculate these confidence
1884880	1887120	intervals and these uncertainties.
1888120	1888840	The way
1888840	1890740	this is done, the way we describe how
1890880	1892920	a biological entity changes over time
1892920	1894800	is, in mathematics,
1894800	1896820	called ordinary differential
1896820	1898920	equation. So this basically
1898920	1900940	describes to you how your biological entity
1900940	1902900	B is changing,
1902900	1904740	so the D represents the change, okay, it's the
1904740	1906580	change of B over
1906580	1908700	the change of time T, okay?
1908700	1911000	How does B change with time?
1911200	1912800	And this is now depending on
1912800	1914940	certain parameters, P, P1
1914940	1916980	and P2. And the plus
1916980	1919040	always means, you know, you get something more,
1919040	1920720	so depending on this parameter,
1920720	1922840	P1, you get more of B if you
1922840	1924840	already have B. So
1924840	1926800	if you have
1927340	1929920	something that can sort of reproduce,
1929920	1931140	like a virus can reproduce
1931140	1932880	or we can reproduce, you know, you
1932880	1935240	will get even more if you have more people to start
1935240	1937260	with, right? And then basically
1937260	1939260	the reproduction rate is your parameter
1939260	1941180	P, and it's positive because you get
1941180	1943080	more. But then
1943080	1945220	this curve of B is not only going up, it's not
1945220	1947200	only just getting more with time, right? It's also
1947200	1949180	going down. So there's also
1949180	1951420	another parameter, P2, and there's a minus
1951420	1953260	because it's getting less, that
1953260	1954280	tells me that
1955600	1957400	B is... the change of B over time
1957400	1959340	is getting negative. There's something
1960260	1961420	subtracted from it.
1961620	1963220	And of course, the more B we have, the more likely
1963220	1965260	it is that, you know, it destructs itself. You know,
1965260	1967240	humanity is very good at destructing itself
1967240	1968580	somehow, unfortunately.
1969560	1971320	But just by itself, it won't go just
1971320	1972740	up and down there. We need
1973320	1975560	another component here that influences this.
1975560	1977280	And this other component is A.
1977740	1979320	So for A, we can actually
1979320	1981460	write down a very similar equation.
1981460	1983300	So how does A change over time? Well,
1983300	1985460	it grows, but not only depending on A
1985460	1987400	but also on B, because, you know,
1987400	1989720	it destructs it, it eats it, for instance.
1989720	1991340	You know, humans are very good at eating
1991340	1993640	stuff, and also other living
1994060	1994780	beings.
1995940	1997480	But everything that lives also has to
1997480	1999460	die. So the more A we
1999460	2001600	have, the more likely it is that it's also dying
2001600	2003460	with another parameter, P4.
2003760	2004900	And then, you know, we get
2005420	2007480	more B, but then A gets more
2007480	2009240	to eat, but then eats more, and then
2009240	2011240	B goes down again. So
2011240	2013260	there's little A that can eat up B
2013260	2015520	and B can re-arrive, and then it's going up and down.
2015520	2017220	So this is sort of how nowadays
2017220	2019320	coupled ordinary differential equations,
2019320	2021340	because there's two of them and they are depending on one
2021340	2023520	another, we can describe the system's
2023520	2025460	behavior and even these weird
2025460	2027580	oscillations, like going up and down.
2028320	2029480	Also from a mathematical
2029480	2031520	point of view, a very interesting
2031520	2033600	observation of this system behavior.
2034360	2035240	But now
2035240	2037240	I talked about A and B a lot, but
2037960	2039140	saying B, I don't
2039140	2041220	necessarily mean a bee. Of course, a bee is also
2041380	2043340	a biological entity, right? The bee that gives
2043340	2045280	you honey and stuff. But B could be
2045280	2047280	anything, honestly. We are talking about
2047280	2049360	the human animal, the marine animal
2049360	2051360	like mice. It could be the number
2051360	2053260	of cells that are changing in our
2053260	2055060	body. It could be bacteria
2055060	2057240	that are also in our body, also cells.
2057240	2059480	It can be this virus, right? This is a
2059480	2061420	picture of a virus. But it
2061420	2063380	can also be just certain molecules that
2063380	2065180	change over time. For instance, here an antibody
2065180	2066360	that is produced against
2067070	2068440	this virus.
2069940	2070540	So
2071200	2073040	these are all options of biological
2073040	2074960	entities that we can describe in biological
2074960	2077300	systems. But to describe them
2077300	2079320	actually we have to think
2079320	2081260	on which level are we
2081260	2082260	talking.
2083300	2085120	Because biology happens on
2085120	2086980	many different scales, in space
2086980	2088920	and in time, right? The human being is
2088920	2090460	on the order of
2091260	2091920	magnitude
2094320	2094800	like
2094800	2097420	one to two meters, let's say.
2098760	2099720	Maybe basketball
2099720	2101460	players are more like two meters
2101460	2103900	and other people more like one meter.
2103900	2104800	Never mind.
2105700	2107820	But then we talked about
2108260	2109660	a mouse that is certainly not a meter
2109660	2111720	but more like centimeters.
2111760	2113500	And then we have cells that you have
2113500	2115320	seen that are more like micrometer
2115320	2117720	which is a meter divided by a million.
2118740	2119060	And then
2119060	2121240	in these cells we have small
2121740	2122220	subunits
2123340	2124420	that are even
2125680	2127340	smaller. Bacteria are also smaller
2127340	2128780	than our body cells.
2129460	2130880	And here we are then in the order of magnitude
2131680	2132160	of
2135780	2137060	a meter divided by
2137340	2139500	a billion, okay? So ten to the minus
2139500	2141800	nine, a nanometer.
2143820	2144340	so... and
2144340	2147320	all these scales, you know, changes are happening all the
2147320	2148980	time. That's how we are being alive.
2148980	2151040	Because molecules are turned over, cells
2151040	2153140	are turned over, our bodies are being turned
2153140	2155080	over. Everything happens there.
2155080	2156100	And this not only happens
2157080	2159180	on a very slow scale, like billions
2159180	2161140	3.6 billion years, this earth
2161140	2163060	does exist and life
2163060	2164340	does exist.
2165140	2167080	But also in a matter of hours,
2167080	2169300	in a matter of days, in a matter of minutes, in a
2169300	2171480	matter of seconds, you know, molecules
2171480	2173820	change their conformation in the order
2173820	2175300	of microseconds. Again, a
2175300	2176840	second divided by one million.
2176840	2178740	And this all, you know,
2178740	2180660	this is all part of this one
2180660	2182620	biological system that we are trying to describe
2182620	2184720	and there's no one out there who can
2184720	2186800	tell me that there's a way to
2186800	2188780	handle this complexity without a computer,
2188780	2190840	okay? So we need computer science
2190840	2192540	to address all this, to understand
2192540	2194660	biology at all these scales.
2195080	2196100	And if you're curious
2196640	2198520	for this, don't you worry
2198520	2200300	because there's a wonderful database at
2200300	2202700	bionumbers.org initiated
2202700	2204120	at Harvard Medical School
2204120	2205800	with you guys in the US
2205800	2208320	where you can ask the question, okay, how big is a cell?
2208320	2210080	How many molecules are in a cell?
2210080	2212160	How does a human body cell
2212160	2214300	is different from a bacterial body cell,
2214300	2216540	from a fruit fly, from a
2216540	2218500	worm? All these numbers are there
2218500	2220520	with references. You can look up the
2220520	2222420	original scientific literature to see
2222420	2224640	the evidence for these numbers. And it's just
2224640	2226400	wonderful looking at
2226400	2228280	all this information that
2228280	2230380	is out there. And not
2230380	2232100	only this information is out there,
2232100	2234500	but it's being continuously generated.
2234500	2236000	We are living in a time now, in an
2236000	2238000	era where biological data is
2238000	2240000	not, for the first time, not a bottleneck
2240000	2242020	anymore. You don't need to spend three
2242020	2243380	years of a PhD to
2244020	2245940	gather one data point of
2245940	2248000	information about one protein.
2248260	2249680	No, we can now
2249680	2251860	sequence our
2251860	2253940	entire genetic information, DNA
2253940	2255980	of a patient upon
2255980	2257860	diagnosis, after treatment, and
2257860	2259880	really understand how our medical
2259880	2260980	treatments affect
2262460	2264000	our biology on so
2264000	2266060	many levels. So we are living in
2266060	2268140	an era of biological
2268140	2270040	big data. But speaking to you, maybe
2270040	2272080	of course I have to say that in fact
2272080	2274460	there's only one true big data.
2274800	2276120	But here I really speak
2276120	2278220	about data
2278220	2279980	and biology that is
2279980	2282120	out there on all of these
2282820	2283980	scales that I've just
2283980	2285980	introduced to you. So
2285980	2287540	data is not a problem anymore,
2287540	2289560	but we need concepts how to
2289560	2291000	work with this data.
2291780	2293700	And in systems biology, there's two
2293700	2295540	sort of complementary
2295540	2297540	approaches to deal with all this
2297540	2299600	data. So again,
2299600	2301940	data can be on the level of organisms,
2301940	2303640	it can be on the level of cells, or it can
2303640	2305220	be on the level of molecules.
2305740	2307460	But the problem is you hardly
2307460	2309740	ever have just one of each of these.
2309740	2311460	So usually you have many
2311460	2313520	human beings in a clinical study, you have
2313520	2316400	many different cells that make up our body,
2316400	2318280	like 30 trillion of them
2318280	2319960	to be precise, or many
2319960	2322120	molecules that do something, right?
2322120	2323780	Many virus particles
2323780	2325860	or anything of that like.
2325980	2328320	So we have two different ways to approach them.
2328320	2329980	Either you do it sort of
2329980	2332100	bottom up, where you say, okay,
2332100	2333560	let's just look at
2334270	2335860	some of these different molecules
2335860	2338080	where we know actually what they are doing,
2338080	2340160	how they interact with one another.
2340820	2342120	And then we try to
2342120	2344240	sort of describe this
2344240	2346060	in a dynamic way,
2346060	2348120	so how they change over time. So first one
2348120	2350160	molecule interacts with another one, and then
2350160	2352880	these two interact with the third one.
2353100	2354320	And then we understand the small
2354320	2356520	part, like we start from the ground,
2356520	2358180	we understand the small part of it,
2358180	2360100	the small module, and then we sort of
2360100	2361440	subsequently try to
2361980	2363700	connect these modules.
2364360	2366100	And as I said, there,
2366100	2368280	based on a lot of knowledge, what is going on,
2368280	2369460	we have some mechanistic
2370880	2372140	understanding how these gear
2372140	2374060	wheels really fit into one another,
2374060	2376020	and then we can see how this system is changing
2376020	2376960	over time.
2377940	2379820	But then there's another way to
2379820	2382120	look at a biological system,
2382120	2384620	let's say a patient, a COVID-19 patient.
2384680	2386220	Just look at them from the
2386220	2388040	top, okay? You don't really know what's going on
2388040	2389980	in this patient, okay? It's a black box,
2389980	2392100	this patient. But now you give a certain
2392100	2394000	drug to this patient, this is
2394000	2396040	your input into the system, into your black
2396040	2398220	box, and you see how the patient reacts
2398220	2400160	to it. And this is your output, your reaction.
2400160	2402120	And then you do this not only with one, but with
2402120	2403780	the entire cohort of
2403780	2406060	COVID-19 patients, and you try
2406060	2408220	to see patterns, okay? Whenever we give
2408220	2410160	this to a certain patient, where maybe we have
2410160	2412220	all these health records,
2412780	2414200	there's always a good outcome.
2414200	2416280	Or maybe not a good outcome. And then you can start
2416280	2418380	to try to understand this input-
2418380	2420120	output relationship. Needless to say,
2420120	2422220	for this you need a lot of data, many patients,
2422220	2423440	many drugs that you
2424280	2425480	potentially test.
2425800	2428280	But then you can learn associations.
2428280	2430340	Okay, smoking
2430340	2432260	history is always connected
2432260	2434400	to a bad outcome, or as we said, a high BMI
2434400	2436140	is always connected to a bad outcome.
2436220	2438220	And these are all these machine learning, artificial
2438220	2439740	intelligence methods.
2440920	2442380	Okay, so these
2442380	2444640	are sort of the two complementaries,
2444640	2446240	not schools of thoughts, but approaches
2446240	2448340	that are not mutually exclusive, right?
2448340	2450200	Because you can, while looking at these
2450200	2452480	patients as a black box, you can still start
2452480	2454300	with some, for instance, cells of
2454300	2456360	the immune system that are attacked by the virus, you can also
2456360	2458320	start with some bottom-up modeling there, and then you
2458320	2460340	can again put it all together.
2461440	2462520	But to give you
2463280	2464440	a precise example,
2464440	2466580	I want to use the bottom-up
2466580	2468560	approach now as an example, okay?
2468560	2470300	So again, where we look at how
2470300	2473000	biological entities change over time.
2473240	2474460	And actually these models,
2474460	2476380	particularly also with these going ups and
2476380	2478180	down, these oscillations,
2478180	2480360	they were introduced a long time
2480360	2481780	ago for
2483300	2483820	for
2485420	2486020	foxes
2486020	2487900	and rabbits,
2488580	2490120	but biology is not just
2490120	2492320	that. And in fact, I wrote a
2494140	2494740	book
2496240	2496840	on
2496840	2498260	rabbit-free
2498260	2500200	biology. However, we can still
2500200	2502440	use all this mathematics that was established
2502440	2504240	to now bring
2504240	2506160	this research back to the future,
2506160	2508120	using this mathematics to understand something
2508120	2510440	about molecular biology.
2510440	2512020	And the last example
2512020	2513740	I want to give you,
2513740	2515960	before I hand over back to Noah,
2515960	2519200	is the example of a
2519200	2520420	molecule that controls
2520420	2522160	the survival of
2522760	2525240	our red blood cells.
2525240	2526320	It has this complicated
2526320	2527980	name, signal transducer and activator
2527980	2532360	of transcription, STAT5.
2532360	2534180	The only thing you need to know about this molecule
2534180	2536380	is it can bind
2536380	2537880	DNA, okay? It can bind our
2537880	2539260	genetic information and then
2540140	2541340	exert certain
2541340	2543400	tasks there. And if
2543400	2545360	we have it, our red blood cells are
2545360	2547360	surviving, and if this
2547360	2548700	molecule is not
2549140	2551660	at our genetic information doing its job,
2551660	2553400	then red blood cells
2553400	2555300	are dying. And so do we
2555300	2557520	at some point. So this molecule
2557520	2559300	basically shows us
2559440	2561280	a life-death decision,
2561280	2563420	which starts with a simple,
2563420	2565320	very simple question. Does 1
2565320	2567280	plus 1 always give 2? Because
2567280	2569160	if we have two of these molecules, they are supposed
2569160	2571280	to form this complex of two molecules.
2571940	2573180	And this sounds simple, but
2573180	2575220	actually biology is more complex than that, because
2575220	2576700	we not only have one
2577120	2579200	family member of this molecule, we have two
2579200	2581520	of them, the blue one and the red one.
2581900	2583640	And also this can be modified.
2583640	2585160	It can be chemically modified to be
2585720	2587260	switched. So this
2587260	2589280	small p here is a chemical group
2589280	2591040	that can be added to this molecule,
2591040	2593200	and then it's switched from the off state
2593820	2595440	by this addition of this chemical
2595440	2596700	to the on state.
2597640	2599780	And so we basically have four different
2600580	2601460	molecules that
2601460	2603400	potentially can do this job of finding
2603400	2605360	genetic information and making sure
2605360	2606860	our blood cells are surviving,
2607320	2609520	which gives us a matrix of potential
2609520	2611960	biological entities here.
2612300	2613640	Not this kind of matrix,
2613640	2615440	but more like this kind of matrix.
2615820	2617380	So from all these molecules, we can
2617380	2619100	form these potential complexes. And now
2619100	2621320	the problem is they are very small, so we cannot just
2621320	2623460	look at them to see which of them do actually exist
2623460	2625320	and what do they do.
2626240	2627380	But we do have
2627380	2628380	experimental data
2629140	2631420	where we see how these complexes change over
2631420	2633680	time. So what we see here
2633680	2635560	on the top is basically we see
2635560	2637460	how many of the switched on molecules
2637460	2639420	we have as compared to all
2639420	2641480	possible molecules. And then
2641480	2643520	we also see how the number of the blue
2643520	2645400	molecules compared to all molecules change
2645400	2647560	over time. And then here we basically
2647560	2649400	see how these two
2649400	2651440	ratios change with
2651440	2652420	respect to one another
2652420	2655140	in these cells.
2655380	2657600	We can measure this, and then with a mathematical
2657600	2659100	model, we can basically describe
2659840	2661200	this data.
2661280	2663160	As you can see, again, we have our
2663160	2665260	simulation and also the confidence
2665260	2667420	interval here. And now with
2667420	2669460	the simulation, we can basically look back and see
2670380	2671640	which kind of molecules
2671640	2673300	from this matrix do actually
2673300	2675440	exist. I mean, which one of those
2675440	2677480	we need to describe this data.
2677680	2679400	And the answer is surprisingly few
2679400	2680660	of them we only need.
2681240	2683420	And this is something we cannot just look at.
2683420	2685520	We really needed the
2685520	2687360	computer to find out what's the probability
2687360	2689020	that these molecules exist.
2689300	2691040	And this we also published. It's not
2691040	2693200	an A work, but again, A and B because
2693200	2695360	these are two molecules. And with this, maybe
2695360	2697400	we better understand now how our
2697400	2699000	red blood cells are surviving and
2699000	2700300	be happy.
2700980	2703020	To wrap this all up,
2703020	2705280	what I've shown to you is basically there are
2705280	2707160	standing parallels between computer viruses
2707160	2709000	and biological viruses.
2709060	2711400	We can apply computational algorithms
2711400	2713380	to address challenges that we are facing
2713380	2715260	in the COVID-19 pandemic.
2715600	2717200	And systems biology helps us to answer
2717200	2719160	complex biological questions with quantitative
2719160	2721260	data, for instance, how
2721260	2723200	molecular switches decide between life
2723200	2725320	and death in our blood.
2725480	2726840	And Noah will give you
2726840	2728260	some more teasers
2729020	2731320	how this all relates.
2731320	2733120	I'm just saying, if you
2733120	2735280	feel intrigued by this,
2735280	2737120	don't hesitate to reach out to us.
2737120	2739220	We will be happy to also
2739220	2740900	share more information.
2740900	2743560	Having said that, I hand over back to Noah.
2749430	2751450	Hi, and welcome back.
2751450	2752290	For anyone that
2754350	2755170	joined us
2755170	2757230	after my introduction, I will just
2757230	2758290	explain again that
2759810	2761050	my name is Noah.
2761050	2762890	I'm a Junior Master Scientist at the Weizmann
2762890	2764650	Institute of Science
2764650	2766270	here in Israel.
2766270	2767990	And my background
2767990	2769190	is from...
2770370	2771830	My background is
2771830	2773830	as most of you guys,
2773830	2776050	I was a reverse engineering and
2776050	2778150	cybersecurity expert at
2778150	2780130	Checkpoint for many years. And when I
2780130	2782050	joined Weizmann, I had to close
2783350	2784030	a huge
2784030	2786010	gap in my knowledge
2786010	2787230	in biology.
2787970	2789410	And I was...
2789410	2791870	What was stunning for me was the parallels
2791870	2793950	that I found so often in my
2793950	2795490	in the material that I
2795490	2797530	studied in the courses and also for the lab
2798570	2799150	with
2799720	2802230	the knowledge that I have from the past.
2802290	2804210	So, I will just give you
2804670	2805710	a few examples of
2805710	2808050	the things we have in biology
2808050	2810090	that will look very
2810090	2812310	familiar for you.
2812490	2813290	So,
2813840	2815990	without any further
2817610	2818570	introduction,
2818570	2819610	let's start with
2819610	2821870	cancer cells. Okay, so cancer
2821870	2823790	cells, unlike
2823790	2826050	most of the people think about them,
2826510	2827770	they are not just cancer
2827770	2829890	that accumulate in our body. This is
2829890	2830850	more like
2832570	2833870	cells in our
2833870	2836070	body that have superpowers. They evolve
2837270	2837870	better
2837870	2839750	than the rest of the cells in the body
2839750	2841790	due to a mutation
2841790	2843970	from a radiation,
2843970	2845890	from a UV light, or from
2845890	2847890	any other, from smoking,
2848870	2849810	or just
2849810	2852150	by chance in the body.
2852150	2853750	And then, what's happening with
2853750	2856010	these cells is that they become selfish.
2856010	2857530	Which means that instead of
2857530	2859230	working together
2859570	2861890	with the rest of the cells in the body,
2861890	2863850	they are selfish enough
2863850	2865490	to try to control
2867150	2867890	and to
2867890	2869870	get all of the resources, for example,
2870370	2871570	they are actually
2871570	2873630	sending signals to the
2873630	2875090	body to build
2875810	2877670	veins in order to get more oxygen
2877670	2880790	into the tumor environment.
2881150	2881410	And
2881410	2882370	one
2884270	2885790	stunning fact that I learned
2885790	2887250	during my courses
2887250	2889810	is that not only that the cancer
2889810	2891790	cells are selfish, they also
2891790	2893730	try to evade the immune
2893730	2895670	system, because
2895670	2897610	one of the roles of the immune system is
2897610	2899470	not only to
2899470	2901490	eliminate pathogens, but also
2901490	2903730	eliminate cells that are selfish, like
2903730	2905670	cancer cells. Because cancer cells, like we
2905670	2907610	have them on a regular basis in our
2907610	2909430	body, and
2909430	2911650	this is one of the main roles of the immune
2911650	2913610	system, to detect
2913610	2915030	them and
2915670	2917190	to eliminate them.
2917470	2919550	So, it's not
2919550	2921530	only that our immune system
2921530	2923670	is working as an antivirus, but
2923670	2925710	the thing here is that the cancer cells
2925710	2927030	are fighting back.
2927410	2929350	So, the picture here is representing
2929350	2931730	the interaction between cancer cells and different types
2931730	2933810	of immune cells. Cancer cells
2933810	2935750	are the one that shows in blue.
2935750	2938090	Immune cells are in different colors, and
2938090	2940130	what they are trying to do
2940830	2941810	is to
2941810	2943890	send... so, you can
2944430	2946650	think about the rest of the
2946650	2947630	colorful cells as
2947630	2949730	different types of antivirus from
2949730	2951950	many different companies,
2951950	2953570	or maybe
2953570	2955770	different models that we have in one
2955770	2957790	antivirus. And what the cancer cell
2957790	2960250	is trying to do is to actually reprogram
2960250	2961810	or to send
2963090	2963610	signals
2965150	2966710	to different immune cells
2966710	2968690	in order to evade
2968690	2970850	them. And this is something
2970850	2972810	that was, you know, when I heard about it,
2972810	2974910	it was really familiar for me. Like,
2974910	2976650	there were many models that tried
2976650	2979590	to evade my VM, to evade
2979590	2980270	my
2981390	2982610	AIDA, my
2982610	2984770	OLADPG, or the antivirus
2984770	2986270	in general. And
2986270	2988870	so, for me, it sounds pretty familiar.
2988870	2989850	So, apparently,
2989850	2991810	it wasn't our
2991810	2993230	idea to evade
2993850	2995770	people, to evade
2996950	2997950	entities that
2997950	2999830	are trying to detect us.
2999830	3001770	And I can promise you that cancer
3001770	3002670	cells are doing it in
3004150	3006190	a much more efficient way.
3006270	3008050	For example, one
3008050	3009650	very stunning
3010430	3011870	example that I have
3012590	3013650	is that
3013650	3015650	every cell is supposed to
3015650	3016970	present
3018730	3019290	a molecule
3019290	3020830	on its surface, which
3020830	3023590	the name of the molecule is an MHC.
3023610	3025270	And this molecule
3025270	3026250	is actually like
3027370	3028490	the info
3028490	3030950	that an executable
3030950	3032570	can have, which
3032570	3034790	describes what
3034790	3037250	these cells are producing,
3037250	3039070	what are the proteins that are running inside
3039070	3041010	of it. And cancer cells,
3041010	3043130	what they actually do is they just
3043130	3044770	don't show, they just don't present
3044770	3046870	these MHC molecules,
3047290	3048570	so it's really hard
3048570	3050890	for the immune cells
3050890	3052890	to understand what actually is going on
3052890	3054150	inside of them.
3055530	3055770	And
3058370	3058970	here
3058970	3060790	you can see something that
3060790	3062710	we call a hot tumor.
3062710	3064790	If a tumor is immunologically
3064790	3066690	hot, it means that the immune system
3066690	3069090	is operating here. However,
3069090	3070890	when cancer progresses, it learns
3070890	3072830	how to deactivate immune cells
3072830	3074690	and make tumors
3074690	3076970	immunologically cold. Like, look at this
3076970	3078890	tumor, and we can
3078890	3080490	see here many different
3080490	3080790	of
3082710	3084930	immune cells that are trying to detect
3084930	3086470	and trying to
3086470	3088830	get attached to it and to understand what's going
3088830	3091050	on inside. But after a
3091050	3093310	while, what we can see in tumors
3093310	3095110	that are actually
3095110	3097110	successful with evading the immune system
3097110	3099690	is that they become immunologically
3099690	3100810	cold tumors. You can see
3100810	3101970	that most of the
3102790	3104570	immune cells just gave up
3104570	3105230	and
3106630	3108510	they said, okay, whatever,
3108510	3110850	I have other stuff to do, and they just left
3110850	3112350	the tumor there alone.
3112690	3113410	This is why
3115690	3116510	you get
3117030	3119150	a tumor in your body.
3119610	3121050	And here is just
3121050	3122770	what you can see here
3122770	3123410	is just
3125550	3126490	a network of
3127090	3129210	different... this is just
3129370	3130190	a brief
3131070	3133170	knowledge that we have so far about
3135770	3136630	all the
3137210	3139330	different ways that cancer
3140150	3141390	has in order
3141390	3143490	to try to evade the immune system.
3143590	3145210	Generally, cells can communicate
3145210	3146530	with each other using diverse
3147210	3149170	signal transaction networks
3149170	3151690	that lead to different outcomes.
3151690	3153070	And what cancer cells do, they
3153070	3155210	upregulate the pathways which
3155210	3156530	inhibit normal
3157350	3158850	function of immune cells.
3158850	3160710	They don't even
3160710	3161610	just
3162930	3163870	tell the
3164970	3166730	specific immune cell, you know, just leave
3166730	3168890	me alone. They also ask him to spread
3168890	3170870	the rumor and tell
3170870	3172650	the rest of the immune cells, you know,
3172650	3174730	this guy, it's okay,
3174730	3176450	you can just leave him alone.
3176910	3178510	And this is something that was
3178510	3180530	really fascinating for me to learn.
3181650	3182590	And here
3182590	3183570	is an example
3184230	3185490	where cancers are
3185490	3187830	overexposing PD-L1,
3187910	3190170	a molecule that suppresses T-cell.
3190170	3191570	Important cell type involved
3191570	3193290	in cancer cell elimination.
3193570	3195630	On the right, you can also see
3195850	3197550	a reprogrammed immune cell called
3197550	3199590	tumor-associated macrophage,
3199590	3200890	which helps to spread
3200890	3203590	the inactivating signal instead of
3203590	3205430	fighting cancer. Okay, so
3205430	3207670	you can see that it's not only telling him, listen,
3207670	3210150	I'm okay, it also asks him,
3210150	3211730	it also asks these cells,
3211730	3213790	you can also tell the rest of the guys
3213790	3216110	that I'm fine and to leave me alone.
3216190	3217830	And this was amazing.
3217830	3219850	I don't know too many models that are
3219850	3221670	actually so
3222410	3223650	powerful and sophisticated
3223650	3226070	and it was amazing for me to learn.
3226870	3227950	And my next
3227950	3229730	example, which is actually one of my
3230770	3232490	favorites, is bacteriophage.
3232830	3233830	Okay, so
3233830	3235730	what is a bacteriophage? Here you
3235730	3237250	can see,
3237250	3239630	let's take a look
3239630	3241090	at this very beautiful
3242670	3243150	gif
3244490	3244970	of
3246390	3247530	like, okay, so
3247530	3249610	I'll just explain. First of all, bacteriophage
3249610	3252090	is like my favorite creature so far.
3252570	3253730	And what it's actually
3253730	3255490	doing, there is a creature
3255490	3257570	that we all know, it's known for bacteria.
3257710	3259530	But there is also another tiny
3259530	3260830	picture and
3262290	3263650	what bacteriophage is actually
3263650	3264990	trying to do is to
3265910	3267310	inject its DNA
3267930	3269270	into the bacteria
3269270	3271450	and then the bacteria, instead of
3272490	3273490	running their
3273490	3275550	own genome, it will run
3275550	3277630	the bacteriophage genome.
3277630	3279570	Does it sound familiar to anyone
3279570	3281570	else? Apparently also with this we didn't
3281570	3283490	come up. And this
3283490	3285290	was amazing for me to understand.
3285290	3287510	You can actually see it. This is actually
3287730	3289190	a bacteria,
3289190	3291650	and here you can see a bacteriophage
3291650	3293230	that are attaching
3293230	3295650	into it and injecting
3295650	3297870	their genome into the
3297870	3299630	bacteria. And here you
3299630	3301630	can see this is a very sad story because the
3301630	3303650	bacteria, sometimes they
3303650	3304510	understand that they are
3306430	3307310	doomed.
3307450	3308930	But this is just a gif.
3309270	3311890	So I will just explain a bit about
3312630	3313750	the progress of it.
3313750	3315630	So what you see here is that there is
3315630	3317590	attachment of the bacteriophage into
3317590	3319030	the bacteria. Bacteria is
3320270	3321430	the bacteria that
3321430	3324070	you know, that we all have in our body.
3324070	3325330	And this is the bacteriophage
3325890	3328790	which has a very, very
3328790	3330050	nice and interesting
3330050	3333050	picture and structure.
3333070	3334110	So what it's doing
3334110	3336970	is actually attaching into the bacteria.
3337030	3338070	It's injecting
3340370	3341850	its own genome
3341850	3342350	into the
3343230	3344130	bacteria's inside
3344130	3346270	surface. And then the bacteria,
3346270	3348050	instead of running, like you see, this is the
3348050	3350150	genome of the bacteria.
3350150	3352410	Instead of running its own code,
3352410	3354530	it's running the bacteriophage code.
3354690	3356470	And then the whole purpose of it
3356470	3356930	is that
3358350	3360270	there are more and more
3360270	3361590	bacteriophages that are
3362650	3365030	being formed inside of the bacteria.
3365190	3365650	And then
3366210	3368590	the poor bacteria, you know,
3368590	3370210	are just full of
3370210	3372870	bacteriophages until they release each other.
3372870	3374590	Until they get, like,
3374590	3376310	it's too much for it.
3376310	3377730	And they just go out
3377730	3379590	and infect even more and more
3380270	3381610	bacteria that are inside
3381610	3382930	of this environment.
3383970	3384110	And
3385650	3386670	another thing
3387210	3389610	that I'm pretty sure that most of you heard
3389610	3391570	about, if you are interested in
3391570	3393470	biology at all, is
3393470	3395710	something that is called CRISPR.
3395710	3397670	So we all know that CRISPR is being
3397670	3399190	used these days in order to
3399750	3401730	kind of engineering the babies
3401730	3403510	of the future and stuff like this. But what was
3403510	3405410	fascinating for me to learn is that
3405410	3407630	actually CRISPR, the whole
3407630	3409290	purpose of it, is
3409290	3411510	this is a defense
3411510	3413570	mechanism of the bacteria
3413570	3415410	in order to detect bacteria,
3415410	3418050	in order to detect phages.
3418050	3419650	So what you see here is
3419650	3420990	that this
3421630	3423630	pink evil thing is
3423630	3425610	the bacteria. And what
3425610	3427610	it's trying to detect, and
3427610	3429650	this is something that we actually
3429650	3431530	stole from them,
3431530	3433510	is what it's trying to detect is
3433510	3435250	the sequence that he also
3435250	3436090	that he
3438290	3439370	saw before
3439370	3441270	of phages that try to
3441270	3443130	attack it. And the way
3443130	3445150	that it's doing this
3445150	3447150	is just, you know, he has a sequence
3447150	3448930	that he remembers
3448930	3451390	and then he just tries to
3451390	3453570	do pairing, like to compare
3453570	3455710	the new
3455710	3457030	DNA that he sees
3457030	3459150	inside of the cell
3459150	3459830	and
3461050	3462870	the sequence that he already knows that
3462870	3464810	this belongs to the phages.
3466350	3474330	So, of course, I think
3474330	3475570	it's very
3476650	3477850	obvious that
3479030	3480390	this Cas9
3480390	3482450	system is very similar to what we know as
3482450	3484290	MD5, or the hashes of the
3484290	3485850	models that we have inside
3487110	3488470	many different
3488470	3489790	antiviruses.
3491070	3491930	And another
3491930	3493910	very cool thing is that it's not
3493910	3496230	only that the bacteria is detecting
3496230	3497870	phages, like
3498610	3499910	using the known
3499910	3501850	sequence, is that the phages
3501850	3504170	are actually invading this defensive
3504170	3506370	system by changing their genome.
3506370	3508070	The same way that malwares are changing
3508070	3510250	their MD5, you know, it's
3510250	3512610	enough to change just one bit.
3512690	3513990	Phages are doing the
3513990	3515010	same in order to
3516490	3518290	like, you know, via evolution
3518290	3519990	in order to being able
3519990	3522050	to inject
3522050	3524190	their code into new bacterias.
3524950	3525470	And
3525990	3527890	so this mechanism that we
3527890	3529730	actually stole from bacterias, we will
3530810	3531930	we are trying
3531930	3533730	these days to use it for genetic
3533730	3535510	engineering, but
3535510	3536670	just so you know,
3536670	3539870	it wasn't our idea, we stole it
3539870	3541910	from a very actually
3541910	3543950	smart creature, apparently.
3544510	3545270	And I will just
3546150	3547850	since our time is up,
3547850	3548850	I will just
3550010	3551370	say briefly what
3551370	3553850	the role of our specific
3553850	3555590	lab. So what you see is
3555590	3558110	so my research
3558110	3559830	is specifically about
3559830	3562170	the microbiome. What is the microbiome
3562170	3563830	is what we are trying
3563830	3565670	to do as you all
3565670	3567870	I don't know if you heard about it, but
3567870	3569790	we are all full of bacterias
3569790	3571850	inside of our gut and we
3571850	3573090	are trying in our lab
3574830	3575870	to understand
3575870	3577950	and maybe in the future
3577950	3579730	also to manipulate the different bacterias
3579730	3581630	that are inside of our gut and all the networks
3581630	3583870	that are inside of it. First of all
3583870	3585710	we try to find the connection between them
3585710	3586850	and different diseases like
3587610	3589650	diabetes and ALS
3589650	3591310	and obesity.
3592690	3593750	And so this is
3594010	3596050	a very new and interesting field
3596050	3597690	and you should know
3597690	3600030	that you all have bacterias
3600030	3601830	all over your hands, your
3601830	3604170	eyes and specifically inside your gut
3604170	3606130	and it's not just that they are there
3606130	3608050	happily not doing anything, they are
3608050	3609790	actually sometimes very crucial
3611710	3613730	in order to keep the
3613730	3615230	homeostasis.
3616310	3617810	So this was just
3618330	3620050	a brief thing
3620050	3621670	about the microbiome, just a brief
3621670	3623910	introduction to the microbiome.
3623970	3626050	Just a few things that we
3626050	3628090	found out so far in our lab is the
3628090	3629990	connection between the bacterias that are inside
3629990	3631990	the gut and the very well known
3631990	3633810	disease that are inside the ALS because
3633810	3636170	the bacterias, they can
3636170	3637890	send signals from the gut
3637890	3639890	into the brain and
3639890	3641430	there is a very
3642130	3643810	clear connection between
3643810	3646010	the dysbiosis that we will see in the
3646010	3646870	gut and
3648590	3650090	the form of this
3650090	3651990	disease. Here you
3651990	3653270	can see many different
3654510	3656230	types of diseases that are linked
3657090	3658110	to dysbiosis
3658110	3659690	in the microbiome.
3659690	3661770	And another very cool
3662890	3663930	subject we are
3663930	3666150	looking into in our lab is diabetes.
3667190	3667690	Just
3667690	3669870	one thing that I learned
3669870	3671650	and it was really interesting
3671650	3672210	and
3675310	3675810	like
3675810	3677650	this is one of my favorite challenges
3677650	3679850	in this field, is that
3679850	3681830	when you have any kind of disease, it's not like
3681830	3683870	that, for example diabetes, if you have
3683870	3685970	too much sugar in your blood,
3685970	3687630	the problem with it
3687630	3688790	wouldn't be
3689450	3691490	just about the fact that you have too much
3691490	3692990	sugar in your veins.
3693390	3695230	Many different diseases,
3695230	3697370	they are like, there is a hip
3697370	3699530	corruption, let's say,
3699530	3701310	in our body, but unfortunately
3701310	3703510	unlike in new operation
3703510	3705190	systems, in our body
3705190	3707670	there is no such thing
3707670	3709450	as different processes that
3709450	3711330	have different
3711970	3713370	classes of memory.
3713370	3715370	So if there is hip corruption
3715370	3717090	in one process,
3717090	3719490	in one system
3719490	3721470	in our body, it affects the rest
3721470	3722830	of it as well.
3722910	3725290	For example, in diabetes, one of the problems
3725290	3727410	that we see is that
3729190	3730810	we lose the
3730810	3732790	homeostasis of
3734570	3735250	also
3736310	3736990	the
3736990	3738570	CA2 inside our
3741070	3742430	cells.
3742470	3743250	And actually
3743250	3744830	we can see that
3744830	3747490	the blood is becoming acidic.
3748110	3750090	So this is just one thing.
3751550	3753230	so, I would just
3753230	3755350	say briefly that I'm really proud
3755350	3757210	really, really proud
3757210	3759250	that I'm taking part
3759250	3760110	of my lab.
3760110	3763230	The very nice person that you see
3763230	3765330	here is my supervisor. I have a professor
3765330	3767070	who is also a supervisor, and he
3767070	3768470	recently used me as
3769570	3771290	a lab rat. What you see
3771290	3773450	here is an insulin
3773450	3774650	sensor that
3775210	3777130	was attached to my shoulder for
3777130	3778770	around three weeks.
3779030	3781610	But I'm, of course, very proud.
3782150	3783210	I wish for
3783210	3785250	myself and for everyone always to be in the side
3785250	3786570	to help
3788170	3789130	patients get
3789370	3791350	a better cure.
3791850	3793330	And this is something that I'm really
3793330	3795210	happy about, that what I'm
3795210	3797110	doing here is actually helping
3797110	3799730	people that are sick.
3799730	3801150	But yes,
3801510	3802470	also,
3803430	3806450	all of us actually do stuff like this in the lab.
3806450	3807530	Also, the reason that
3807530	3809330	this guy is dressed like this is because
3809330	3811530	during the corona,
3811530	3813530	our whole lab became
3813530	3814690	of course,
3814690	3816850	we are not
3817750	3819570	doing anything that is linked
3820290	3820850	to
3822630	3823750	corona.
3823830	3825070	We didn't do anything
3825470	3827850	about it like before,
3827850	3828870	but during
3829350	3831850	this breakdown
3831850	3832970	of...