Technical Challenges in Synthetic Biology

Top

00:00:11.16 Hi, my name is Vivek Mutalik. I'm a scientist at the Physical
00:00:15.00 Biosciences Division at Lawrence Berkeley National Lab
00:00:17.22 at Berkeley, California. Today I'm going to talk to you about technical
00:00:22.03 challenges in the synthetic biology field. So let's start
00:00:25.05 by defining the synthetic biology as a field. It is a design
00:00:30.13 and construction of new biological systems and parts.And also, redesigning the existing biological systems
00:00:36.12 for useful purposes. Here the main idea is to use
00:00:40.00 computer-assisted design approaches and engineering
00:00:42.18 approaches with biological research. Essentially, engineering
00:00:47.06 biological functions to make them easier, efficient, reliable,
00:00:50.24 predictable, and safe. As a field, synthetic biology
00:00:56.14 has promised a lot of solutions for a variety of problems
00:01:01.00 we are having today, either as an energy, environment,
00:01:03.24 agriculture, health, chemicals, food, and also the
00:01:09.18 understanding of biological complexity -- synthetic biology has
00:01:14.00 promised solutions. As a field, synthetic biology has applications
00:01:19.08 in a wide variety of industries, whether it's energy, environment,
00:01:22.12 agriculture, health, and chemicals. As you can see, there are
00:01:27.07 a variety of applications and each of them have their unique
00:01:30.13 challenges to overcome. Today I'm going to talk to you about
00:01:33.02 a few of these challenges which can be generalized across
00:01:36.22 these different applications. So researchers have developed
00:01:41.12 a variety of approaches to take gene clusters from one organism
00:01:45.17 and put it in a different organism to express it, and produce whatever
00:01:50.19 chemical they are interested in. They have also designed
00:01:53.09 and executed different genetic circuits. One idea of pathways
00:01:58.02 that have been built over the decades to produce these particular
00:02:01.18 interesting chemicals, in this case it is artemisinin production,
00:02:04.17 which is an antimalarial drug. Or as you know, there is a
00:02:08.01 iGEM competition, which is a worldwide phenomenon.
00:02:11.08 Where undergraduates from a variety of different countries
00:02:15.09 participate where they come up with innovation and new ideas
00:02:18.14 and using synthetic biology, they create different solutions
00:02:22.11 for different problems. Now over the years, people have
00:02:27.02 used a variety of approaches where they have
00:02:30.08 developed sophisticated genetic circuits where there is
00:02:33.20 dynamic feedback regulation, where a particular substrate
00:02:36.23 is producing a metabolite which is toxic to the cell
00:02:41.11 producing the product, so the metabolite can be used
00:02:44.02 for giving the feedback to the upstream pathway. It represses
00:02:48.23 the upstream pathway and activates the downstream
00:02:50.20 pathway, thereby reducing the toxicity. Also in Drew Endy's lab,
00:02:55.08 they have refactored bacteriophage T7 with a 12kb
00:02:58.12 replaced with an engineered one. From Chris Voigt's lab, there
00:03:01.15 is a whole refactoring of a nitrogenase cluster, where they have developed
00:03:05.18 more than 500 gene clusters by going through to design,
00:03:09.01 build, and test a learn cycle. Along with that, we have already
00:03:13.22 seen that there's a whole synthetic chromosome synthesized
00:03:18.05 in yeast. And also as you can see from here, there is from
00:03:22.03 the Craig Venter Institute, that this is a 1 megabase DNA
00:03:25.24 from synthetic approaches. Other than these special cases,
00:03:31.10 other than that, all the state of the art has been slow,
00:03:34.10 expensive, and unpredictable. Usually, the way this
00:03:38.07 system has been working is that PhD level scientists
00:03:40.21 plan and execute the material, and they test them, and
00:03:45.05 adapt them, and manually assemble them. Usually
00:03:48.21 as an ad hoc approach, where the application is defined
00:03:51.15 and then they couple the design and fabrication
00:03:54.07 and go through the iterative cycle, and usually,
00:03:58.04 we don't keep track of the mistakes made or the errors
00:04:01.00 done. So we don't know what the lessons are that we learn
00:04:05.07 from every cycle. So as you can see if you zoom out
00:04:09.12 in view, every application has its unique challenges and
00:04:13.03 it needs a lot of money, time, and labor to optimize, because
00:04:15.10 it's not a simple solution. So in this case, a bioprocess
00:04:19.11 platform has multiple branches. It has got substrates
00:04:23.12 that need to be engineered, product needs to be optimized,
00:04:25.15 and the actual reaction system, conditions need to be optimized
00:04:30.01 and so on and so forth. So there's a multifactorial problem
00:04:33.24 we need to overcome. And gene expression is one of the simplest
00:04:37.10 challenges that is involved in this process. So if we give you
00:04:42.15 a simple test to do, to solve with the state of the art
00:04:45.10 methodology that we have today. Take your time to calculate
00:04:49.08 how much money, time, and labor is needed to do
00:04:52.08 all of these simple projects. Even though they look simple,
00:04:55.04 they have whole challenges of making them efficient and
00:04:59.12 selective. So for example, can you engineer a pathway where you use
00:05:04.05 200 grams of isobutanol in a model organism in
00:05:07.21 shake flasks? What if I want to refactor a gene cluster
00:05:11.17 identified in organism-X and express it in a model system
00:05:15.10 -- a model host which produces an anticancer compound Y?
00:05:18.18 And the third challenge will be, can you engineer a phage
00:05:22.14 which can diagnose the pathogenic microbe which is
00:05:26.07 drug resistant. And then finally, what efforts do we need to do
00:05:31.07 to design a bacterial system or virus for cancer treatment?
00:05:35.08 And the answer is, we are not really sure how much
00:05:37.24 money, time, and labor is needed to do all of these
00:05:40.10 things. And people have realized already that the challenge
00:05:45.17 is not only the DNA synthesis, which is getting cheaper and cheaper,
00:05:50.00 though we have increase capability, our ability to
00:05:53.08 design and put things together hasn't gotten better.
00:05:57.17 So right now, we have an ad hoc genetic engineering
00:06:00.23 as I mentioned. We have limited well-characterized
00:06:03.15 parts, and there is no reliability and predictability associated
00:06:08.00 with the parts. We don't know how to put these things together
00:06:11.04 such that they're functional. And then finally, also when
00:06:15.18 we put these parts together into circuits, we are not really
00:06:19.04 sure what are the emergent properties that come up with
00:06:21.06 these complex systems. And then context issue matters,
00:06:26.12 where the genetic circuit is, what are the cellular components,
00:06:29.16 what are the next regions of the DNA. And what happens is
00:06:32.21 that finally, we are not really learning anything from our failures.
00:06:35.24 We don't keep track of the failures and we don't know what is
00:06:38.17 happening. And then finally, as you know, we don't know enough
00:06:42.12 biology to make synthetic biology a predictable engineering
00:06:46.15 discipline. So how do we really accept all of these challenges?
00:06:50.06 And leap ahead? What do we need to do? So as a field,
00:06:53.20 people have already proposed many of these things. This is
00:06:57.14 Drew Endy's very famous review in Nature, which particularly
00:07:02.10 shows three things, which three steps we need to do
00:07:05.03 so that we can start solving genetic engineering to the next
00:07:08.18 level. So for example, we need to decouple the design and
00:07:12.12 fabrication. We need to standardize parts and part junctions.
00:07:15.17 And we need to abstract a genetic function, so abstraction
00:07:19.04 in terms of whether it is a DNA, parts, devices, or systems
00:07:25.02 -- we need to know everything in terms of DNA sequence
00:07:27.18 so we can abstract up to a level where we are working
00:07:30.24 and simplify the problem. So if you see in a simple way,
00:07:34.18 if I define an application, the question should be, how do I
00:07:38.12 design it? What specifications are necessary?
00:07:41.05 How do I model it in a computer so that I know what I'm
00:07:44.16 designing? And then choosing the parts, and then building
00:07:48.05 them. After that, we characterize it and then we debug.
00:07:51.07 And based on that, we can actually find out which step we need to debug.
00:07:55.22 As you can see, the design, build, test cycle that I'm mentioning
00:07:59.02 here is pretty much the same for all of the things.
00:08:02.10 Whether it can be a software for synthetic biology, or it
00:08:04.14 can be a host system if I wanted to engineer a host system,
00:08:07.23 what do I need to do? And then how do I do a design, build, test, learn
00:08:11.10 cycle for a device or system that we want to engineer?
00:08:14.17 Or it can be a DNA part if I'm engineering a promoter, I need
00:08:18.24 to know how do I design it, how do I build it, and test, and learn.
00:08:22.10 So this cycle of iterations, it applies to all of the branches
00:08:27.02 of synthetic biology. So I'll give you a quick primer on
00:08:33.00 regulation of gene expression before we go ahead. Here,
00:08:37.02 RNA Polymerase binds to a promoter in E. coli and then
00:08:40.08 since it's a polycistronic mRNA, it produces all of these three mRNA.
00:08:46.14 Once the mRNA is produced, ribosomes bind to the ribosome binding site
00:08:49.15 and it produces these proteins. As you can see, each of these processes
00:08:53.19 are multi step processes. So for example, in transcription, we have initiation,
00:08:57.11 elongation, and termination. And all of these three things are regulated by
00:09:02.02 multiple different motifs. Same is for translation, it goes through multiple
00:09:07.05 steps, initiation, elongation, and termination. And each of these steps
00:09:10.24 are regulated by multiple different factors. So as you can see,
00:09:14.07 if I want to optimize the protein production, I need to
00:09:17.03 read about all of these factors, along with mRNA degradation.
00:09:19.22 So you can see that if I want to really turn the knobs on gene
00:09:25.09 expression, I have a variety of solutions I can approach. So I
00:09:30.10 can change the copy number of the DNA, I can use promoters,
00:09:33.21 sigma factors, RNA polymerases, and so on. I can use
00:09:37.24 a variety of these new knobs such that I can tune up the
00:09:41.00 protein expression. So as protein expression is a huge complex
00:09:46.03 system, protein expression is regulated pretty much by
00:09:49.12 everything in the cell. So it can be a whole transcription machinery
00:09:54.09 along with the translation machinery, all regulating protein expression.
00:09:58.01 It also gets impacted by the conditions. Replication of the cell,
00:10:02.09 variety of the DNA, and also the growth rate, the degradation of mRNA
00:10:08.00 and proteins, folding factors and so on. So how do we really
00:10:12.02 approach or improve the functional composition within all of this
00:10:15.22 coupling and complexity of varying cellular complex? So, this is a slide
00:10:23.00 from Adam Arkin at LBNL, where the interesting idea would be
00:10:26.05 if I take this expression unit, or cassette, where I'm producing a protein
00:10:30.00 and if I put it across different sites on the chromosome, how
00:10:32.24 does it impact the gene expression? How does it impact it if I take
00:10:36.10 the same gene expression and put it on the single copy
00:10:38.08 vector or a multi-copy vector? And what happens if the same
00:10:42.16 things get impacted by the environmental context? Whether I grow
00:10:45.18 cells in a test tube or I grow them in a shake flask, how does it
00:10:51.10 impact in a reactor? If I put the same expression unit in a
00:10:56.01 different strain, how does it impact the gene expression? So how
00:10:58.19 do we take care of this all context reliability matters into when we
00:11:03.08 are measuring a subunit? So the early methods where we started
00:11:07.05 to assemble parts, this is a registry of standard biological parts, which
00:11:12.22 started to catalog parts from IGEM teams. This is joint biology institute
00:11:17.01 datasheet where they have a plasmid library, and they characterize
00:11:21.02 it in different conditions. This is a similar library from the European
00:11:26.08 group, where they have a library of vectors. And this is the datasheet from
00:11:31.02 Drew and his lab, where they have shown how to really generate a
00:11:34.22 datasheet of parts under different conditions. Such that when we read this
00:11:38.23 datasheet, we will know what happens with the part and how does it
00:11:42.01 behave in different contexts. But now we're aware of a variety of
00:11:47.00 approaches people have put together that are ligation independent
00:11:51.09 methods, restriction site independent methods. But let's imagine
00:11:54.15 if you put together all of these into a circuit, there are different
00:11:57.20 failure modes that can happen which impact the performance.
00:12:01.20 So for example, now we know that promoter impacts like promoter
00:12:08.07 context are very important, it can impact the protein expression
00:12:11.00 by changing the promoter strand. So we need to put in
00:12:13.21 an insulator, but we don't know what the insulator region is.
00:12:18.12 We don't know what is the RBS context, what needs to be in the RBS.
00:12:22.00 What happens in between these junctions between these parts?
00:12:26.03 And what happens if there are scar issues from the restriction enzymes
00:12:30.10 that are between these parts? How do they impact the overall gene
00:12:33.19 expression? We also need to worry about cross-talk, if there are some
00:12:37.21 parts which interact and which do not interact. And then also we need
00:12:42.04 to read about input/output, dynamic range between different gates.
00:12:45.09 And then transcriptional read throughs, so maybe the RNA polymerase
00:12:48.16 does not terminate at the terminator, it reads to a downstream promoter.
00:12:53.04 And then finally, we need to understand how we generate diversity
00:12:59.06 between the genetic circuits. And if we generate diversity, how do we
00:13:02.15 really screen them or measure them. And then finally, how do we
00:13:06.21 take all of this big chunk of DNA and design structures, how do we
00:13:10.20 put them in a particular chromosome or genomic context? And then how do
00:13:16.00 we really quantify using in silico methods? How do we model them,
00:13:20.00 how do we approach these solutions? So one of the approaches people have
00:13:25.21 come up with, particularly in bio-fab, where we have a promoter
00:13:29.22 and then downstream UTR driving the gene expression of gene of interest.
00:13:34.19 And these particular expression units are selected by two different
00:13:40.01 insulator regions, one is upstream insulator and the downstream insulator.
00:13:44.11 So the upstream insulator is made up of a terminator and some
00:13:48.01 regions which we call insulators, and the downstream is another terminator.
00:13:52.12 So the idea is that there is no traffic inside going inside the
00:13:57.02 transcription unit, and nothing is coming out of this unit. So the
00:14:01.18 ideas behind many of these publications that have come up
00:14:05.24 in recent years have really characterized the parts and the
00:14:10.01 junctions. So in this case, there is a promoter library, a variety of
00:14:13.00 methods have been used to generate promoters for E. coli.
00:14:15.11 There are terminators, library of terminators, which are available for
00:14:20.02 E. coli. And also, there are junctions, particularly in this case,
00:14:23.01 Chris Voigt's lab has published this paper where we have a ribozyme
00:14:26.20 between a promoter and then the UTR, where it can cut down
00:14:31.13 the RNA, such that there is no interference between these
00:14:34.13 parts. Also bio fab has shown that this bicistronic design,
00:14:38.18 where there is an upstream RBS and a downstream RBS
00:14:42.16 and then a small peptide coding between the upstream region of interest
00:14:46.21 where if there is any hairpin formation between the RBS and
00:14:50.20 the gene of interest, then it will be removed by the helicase activity of the ribozyme.
00:14:54.16 So it's been shown that using these approaches, if you have a different
00:14:58.17 gene, let's say in this case, GFP and RFP, you get a pretty linear
00:15:02.12 dependency. The rank order is maintained across different
00:15:06.18 contexts. Also, there is a study here shown on the extreme right
00:15:12.15 where the 5' and 3' UTR engineering has been down to where there is no
00:15:16.24 dependency between these parts. And then the most important
00:15:21.05 and also unanswered question here is the optimal design
00:15:23.15 conundrum, where we don't know how do we codon optimize if we
00:15:28.06 are expressing a gene from a different organism in E. coli or any other
00:15:32.11 part, for example. So we really need to spend more time and investment
00:15:36.14 to understand how the gene design is happening and what is
00:15:39.24 the optimum solution. There are some biophysical methods, for example,
00:15:44.05 in this case, RBS calculator from Salis lab has shown that
00:15:47.12 this RBS calculator can be used to verify a biophysical method
00:15:50.02 for expression optimization in a bigger circuit. So what are the things that
00:15:56.09 we already not think of is the host and environmental context.
00:15:59.19 So let's imagine this is my construct that I really want to
00:16:04.00 produce whatever I want to produce or doing the program
00:16:07.11 that I want to engineer. We forget that it's in the context of a
00:16:12.02 whole cellular genetic material and genetic program that already
00:16:15.06 exists. And also we forget that there are the metabolic circuits, which
00:16:19.04 it is a part of. So when we think about the context and host
00:16:23.01 context, particularly, we do not take care of this issue. So what happens
00:16:27.19 with that is that because of these dependencies, this circuit
00:16:30.20 might be using gene expression machinery from the cell,
00:16:33.24 it might be titrating away the protein degradation machinery,
00:16:36.11 it is using too many metabolites and cofactors, impacted by
00:16:42.14 the replication machinery, and so on. And because of that, cells
00:16:45.01 experience a burden exercised on them. And because of the burden,
00:16:50.12 the cell goes through this toxicity response or stress response,
00:16:54.18 and it impacts the growth rate and the circuit performance
00:16:58.06 is impacted. So there are a variety of methods that people
00:17:01.20 have come up with, and I'm going to show you two different
00:17:04.16 methods. One is the relative promoter units from Drew and his
00:17:07.21 lab, where they have taken these two different promoters they sent
00:17:13.02 to different labs and they tested the performance of those.
00:17:15.08 As you can see here, the one promoter (white), it shows a different
00:17:20.00 performance across the labs, and the same thing, the second gray
00:17:24.16 bars are showing a different promoter. As you can see, each
00:17:28.05 same promoter shows a different performance under same conditions
00:17:32.18 in different labs. Obviously, how do we really normalize this?
00:17:36.02 So one idea is to use relative promoter units, where they are dividing one
00:17:40.14 of the activities with the another promoter activity by reducing the context
00:17:45.21 variability. And then the second solution is to come up with a parts score,
00:17:50.18 how we really score a part across different conditions. And this is the
00:17:54.07 work from Bio Fab, where we took several different promoters and
00:17:58.19 UTRs driving GFP or RFP on a medium copy plasmid. So when we measured
00:18:03.15 the fluorescence of two proteins, we see that we can only explain
00:18:08.12 about 40% of the data. So if you use a simple factor, we can come up
00:18:13.13 and score these parts across different conditions. So I'm showing here
00:18:16.21 part scores across different promoters and UTRs, and the error bars
00:18:22.03 are showing how does the particular part vary with respect with its
00:18:27.02 upstream genetic context or downstream genetic context. So you can
00:18:30.17 imagine having this kind of database where the part scores help
00:18:34.06 in designing and choosing parts which are appropriate and reliable
00:18:38.14 for your application across different systems. So the second
00:18:44.13 thing is, how do we really debug the failure modes that we are having
00:18:47.24 with the host engineering. So there are a variety of methods that people
00:18:51.19 have used and people have used a lot of genetic approaches, genetic
00:18:53.22 engineering approaches. And some of them are shown here: MAGE, TRMR,
00:18:58.14 and Cas9 systems have been really useful in E. coli. And not only E. coli,
00:19:03.17 Cas9 has been shown across different systems. So in this case,
00:19:09.05 what happens is that you can create mutations or insert a particular gene
00:19:12.23 where you can have multiple levels of genome engineering scales.
00:19:16.12 And by that, you can reduce the toxicity and effects on the growth rate
00:19:20.15 and physiology. So when there are also other methods, you can also
00:19:25.00 quantify it -- how do we really quantify the stress levels? And there are methods
00:19:28.14 people have used by not only quantifying but also using feedback
00:19:33.04 regulation across the circuits so you can sense the stress and then
00:19:35.20 you can really feedback into your circuit such that it performs
00:19:38.24 with respect to its stress, so that it should not impact the
00:19:43.00 cellular physiology. So one of the examples is from Keasling's group,
00:19:45.24 where they have shown that acetyl-CoA can be converted to
00:19:50.12 mevalonate, which converts to FPP which is toxic to the cell.
00:19:53.05 And FPP goes into amorphadiene. So you can imagine
00:19:57.16 they found out that if you can remove FPP from the larger
00:20:01.24 pool, you can reduce the stress levels. So they found out the
00:20:05.18 FPP response promoters and they used that to downregulate
00:20:11.22 the upstream promoter of the pathway and activating the
00:20:16.16 downstream promoter, by that way, you can remove the FPP
00:20:20.24 toxicity level to a much, much better physiologically tolerant levels
00:20:26.01 in the cell. And the second one is from Tom Ellis' lab, what they
00:20:31.03 did was they have a stress detection methodology, where they have a GFP
00:20:38.06 cloned in on the chromosome which can quantify the stress levels
00:20:41.04 of the protein that you want to express. By quantifying, you can find
00:20:45.11 solutions to remediate and debug the stress levels. So as we move into
00:20:49.05 biology, researchers have started thinking about how do we contain
00:20:53.00 the engineered organisms from being released into the environment.
00:20:57.20 So biocontainment is one of the major challenges that the research
00:21:01.23 community has been thinking about. How do we really safeguard?
00:21:05.11 And the variety of strategies that people have used over the years:
00:21:08.00 One of them is an engineered auxotrophy, in this case, toxicity
00:21:12.21 of that particular system can be engineered so that when it's released
00:21:18.04 into the cell, the cell doesn't survive because of a toxic gene expression.
00:21:22.13 The second method is that you have the deletion of an essential gene
00:21:26.15 and are supplying the amino acid, and the cell will survive in the container
00:21:31.18 but when it's released into the outside environment, it doesn't
00:21:36.21 get this essential amino acid and the cell doesn't survive. And then the
00:21:41.10 third one is an induced lethality, where what happens is when a particular
00:21:45.19 cell gets released into the environment, in the presence of an inducer,
00:21:49.15 you have a gene expression which is toxic to the cell and the
00:21:53.06 cell's survival is compromised. And then finally, gene flow
00:21:57.15 prevention is when a particular plasmid is toxic to a different cell
00:22:07.03 and the gene expression will kill the cells. There are really
00:22:12.00 new methods coming out, which are multiplexing. These are recent papers
00:22:17.02 this is from the Church lab, and particularly, they recoded the E. coli
00:22:21.17 so that all the UAG codons are now UAA, and now this
00:22:25.13 recoded organism, E. coli, can use this synthetic amino acid.
00:22:38.15 So what happens is that in order for cells to survive, you need
00:22:41.12 this synthetic amino acid to be supplied. So in absence of this in the
00:22:46.21 natural environment, the cell doesn't survive. And there are very few
00:22:50.08 escape mutants because of this mutation, this UAG incorporation
00:22:54.11 has been done on multiple essential genes. Now the second approach
00:22:59.03 is where they have two essential histone genes that are regulated
00:23:04.24 by p-gal promoters, which are activated by GEV protein in the presence
00:23:10.11 of an inducer. So in absence of inducer, these two essential proteins
00:23:13.22 are not found and the doesn't survive in the natural environment.
00:23:18.23 And the second level of redundancy they have incorporated is
00:23:22.07 they have induction of a CRE that will cause the recombination
00:23:26.17 between the Lox-p sites and then you lose one of the genes and
00:23:30.08 the cell is incapable of surviving in the natural environment.
00:23:34.15 So if you think about it, we need all of the different tools of
00:23:38.23 wet lab tools and also the software methodology to handle all of
00:23:43.07 the different challenges that we have. So defining how do we
00:23:47.01 design it, and what are the tools to design, we need software.
00:23:49.08 And we need methodology to model approaches and applications,
00:23:53.16 we need software. And along the way, we need different software
00:23:58.02 platforms. So I have listed here some of them, for example, we need
00:24:01.03 design, design and statistical approaches, how do we really use
00:24:05.02 design of experiments. Methodologies, how do we store metadata,
00:24:10.10 how do we start part data, and then the sample tracking and strain
00:24:14.18 location and so on and so forth. There is a lot of open applications
00:24:21.05 for software development for synthetic biology. So this is shown in
00:24:25.21 one of these reviews, where let's imagine that I have an application
00:24:30.02 definition, we have a specification, and based on that, we design
00:24:33.13 the parts, design the system, we can define them using the equations,
00:24:39.23 we can solve them and predict on the computer how they behave,
00:24:43.12 based on that, we can choose parts based on different databases
00:24:47.21 that we have, define them and convert them into sequences, and
00:24:52.14 then assemble them in silico and go into the wet lab and build them
00:24:57.03 in reality. Or maybe order it through the synthesis companies.
00:25:00.21 And then check the experimental data with the in silico predictions.
00:25:05.07 There are a variety of software people have been developing, but
00:25:08.24 there is no one-stop shop for all of these tools together. We are
00:25:12.18 still limited by how do we really keep the samples tracked,
00:25:15.24 what are the failures, how do we keep track of the failures? So we
00:25:19.11 need a lot of improvements on how we handle data. We also need
00:25:24.16 to move away from simple data tables, so all the researchers still use
00:25:29.07 Excel sheets to check the data or analyze the sequences, or analyze
00:25:35.24 the data sets that they have. We need to move away from that to more
00:25:39.00 design and sequence constructions, where design is explained
00:25:43.04 and we have the data along with the designs. We also need
00:25:46.18 more visualization. How do we really access the data and visualize
00:25:50.18 the data. How users are going to use it, score it, or are more user friendly
00:25:56.23 and a nice experience for using the data. So also, if you zoom out
00:26:04.01 from all of these, there are multiple registries coming up across
00:26:07.15 the world. And we need to really come up with a web of registries
00:26:11.04 so that all of these are connected. Such that all of these data sets
00:26:15.09 are shareable and people are able to share the data between the
00:26:20.05 different labs. And research groups are learning from each
00:26:23.14 of these, they're sharing their standards. And so on and so forth.
00:26:26.17 So we need a set of web registries where all the data is accumulated
00:26:31.04 in one place, and it can be shared across the labs.
00:26:34.21 But all of these solutions, right now if you think about all the publications
00:26:38.04 in synthetic biology, they are focused on E. coli or yeast.
00:26:43.21 But we know from industrial biotechnology, there are different
00:26:48.00 organisms being used in industry, not only E. coli and yeast.
00:26:51.24 There are a lot of organisms that have a huge potential
00:26:54.15 for either as bioremediation or production of chemicals
00:26:59.10 or antibiotics. So unless we develop tools for all of the
00:27:03.00 microbes, I don't think we'll be reaching the real potential
00:27:06.03 of synthetic biology. I hope I have given you a brief overview
00:27:10.11 of the challenges faced in at least some generalized way. And
00:27:16.02 I enjoyed talking to you. Thank you.

This Talk

Speaker: Vivek Mutalik

Audience:

Researcher

Recorded: June 2015

Talk Overview

Dr. Vivek Mutalik highlights current challenges in synthetic biology and explains some of the solutions being implemented to address them. Mutalik discusses some of the elements that make current approaches to synthetic biology unpredictable and expensive, and reviews possible ways to move the field forward, including the development of standardized parts with predictable behaviors, robust methods of biocontainment and software that allows data sharing and visualization.

Speaker Bio

Vivek Mutalik

Vivek Mutalik is a staff scientist at the Environmental Genomics and Systems Biology Division and Biological Systems and Engineering Division at the Lawrence Berkeley National Laboratory. Previously Dr. Mutalik was a team leader at the BIOFAB, the first biological design-build-test facility, which focuses on developing extensively tested standard biological parts to facilitate easier engineering of… Continue Reading