Genomics and Cell Biology of the Apicomplexa

Transcript of Part 5: Designing and Mining Pathogen Genome Databases: From Genes to Drugs and Vaccines III

00:00:04.23	I hope you found our tour of
00:00:08.12	computational approaches for identifying candidate vaccine targets
00:00:12.27	of interest.
00:00:15.15	And needless to say, there are many other questions,
00:00:18.08	many other approaches,
00:00:20.09	many other strategies that one might want to explore
00:00:23.06	for addressing problems of interest in malaria biology --
00:00:29.29	questions related to the cell biology of these organisms,
00:00:33.05	questions related to their biochemistry.
00:00:35.28	But in the applied realm,
00:00:38.07	certainly at least a co-equal
00:00:44.05	with the identification of vaccine targets
00:00:46.13	is the identification of drug targets as well,
00:00:49.02	and I'd like to take just a moment to put into context
00:00:52.11	the severe challenges and the reasons why
00:00:55.12	we're so concerned about the identification
00:00:57.15	of new drug targets.
00:00:59.19	If we step back a couple of generations,
00:01:01.22	we were fortunate in having several
00:01:07.13	inexpensive, effective antimalarials,
00:01:11.12	which, there was some hope, would be...
00:01:14.23	would enable us to eliminate
00:01:18.18	this devastating disease from Earth.
00:01:23.20	But unfortunately, the emergence and spread
00:01:27.18	of drug resistance to chloroquine
00:01:30.00	and to antifolates
00:01:32.11	has left us with far fewer effective drugs
00:01:36.16	than we once had.
00:01:38.28	Indeed, the first line course of treatment, now,
00:01:42.09	for malaria is primarily based
00:01:45.24	on a Chinese herbal remedy, artemisinin,
00:01:50.00	which is now... which is now extracted from plants
00:01:54.02	and can also be made synthetically.
00:01:56.19	And artemisinin common... combination chemotherapy
00:02:00.01	is the basis of treatment,
00:02:02.12	but also the source of some anxiety,
00:02:04.16	because history has shown
00:02:07.22	that time and again
00:02:10.23	overreliance on a single drug or a single treatment strategy
00:02:13.13	is likely to give rise to the emergence
00:02:16.04	of drug resistance.
00:02:17.21	And that has stimulated the development
00:02:22.12	of many efforts around the globe
00:02:24.19	to identify new antimalarials.
00:02:27.07	It's estimated that we will need
00:02:29.14	new antimalarials every 5-10 years
00:02:34.00	in order to keep pace with this devastating disease.
00:02:38.24	The Medicines for Malaria Venture
00:02:41.15	is a public-private partnership,
00:02:43.18	an interesting exploration in the development
00:02:46.07	of a sort of virtual pharmaceutical company driv...
00:02:48.23	pulling together the best of expertise in academics,
00:02:52.13	in the biotechnology sector and in the academic sector,
00:02:55.05	with government laboratories and private funding agencies
00:02:59.13	to try to expedite the development of drugs.
00:03:02.23	And I thought it might be informative
00:03:05.04	to look at the Medicines for Malaria Venture pipeline,
00:03:07.19	a pipeline... this sort of representation of drugs
00:03:13.24	is standard in the pharmaceutical industry,
00:03:16.14	describing the earliest stages of drug discovery,
00:03:18.23	from laboratory research through preclinical development,
00:03:23.21	and moving out into... into the clinic.
00:03:25.28	And the Medicines for Malaria Venture drug pipeline,
00:03:28.14	circa 2006,
00:03:32.01	included quite a full portfolio,
00:03:34.11	including a variety of artemisinin-based compounds,
00:03:38.13	indicated in red here,
00:03:41.03	in the late stages of development;
00:03:44.09	several candidate antimalarials
00:03:48.00	being developed using more traditional pharmaceutical chemistry approaches,
00:03:52.07	those indicated in orange;
00:03:54.10	dicationic molecules;
00:03:56.16	or some synthetic compounds that bear similarity to artemisinins;
00:03:59.11	some compounds that are based on natural products,
00:04:02.14	on herbal remedies;
00:04:04.25	manzamine alkalines, here, derived,
00:04:07.02	as it happens, from a sea sponge;
00:04:09.24	other compounds that that one might think of as new drugs from old...
00:04:13.16	based on old strategies,
00:04:15.25	compounds similar to chloroquine or to antifolates
00:04:19.00	that are indicated in pink.
00:04:20.22	But what's interesting is that all of the new compounds
00:04:22.24	entering into this pipeline
00:04:25.07	are based on genomic approaches,
00:04:27.27	and, just as we've discussed,
00:04:30.02	the kinds of strategies that one might want to consider
00:04:32.28	for mining genome databases
00:04:35.12	to identify candidate targets for vaccine development...
00:04:38.12	we could develop similar approaches for drug development,
00:04:42.15	and I encourage you to think about ways of doing that
00:04:45.16	and explore it yourself online
00:04:48.06	using the Plasmodium genome database
00:04:50.20	or other resources.
00:04:52.27	But the challenge is really highlighted
00:04:55.17	by following not just the static view of the portfolio
00:04:58.23	of the Medicines for Malaria Venture,
00:05:00.26	but by considering what's happened in the interim,
00:05:03.03	in the past couple of years
00:05:05.08	since this portfolio was put together.
00:05:07.08	Fortunately, many of these compounds
00:05:10.04	are now emerging out into the clinic,
00:05:12.22	and we can see them passing on
00:05:15.01	into phase III trials.
00:05:18.16	Some compounds have encountered roadblocks
00:05:20.20	and have moved backwards
00:05:22.18	-- hopefully a temporary reverse
00:05:24.18	in which problems encountered
00:05:26.20	with a synthetic peroxide
00:05:28.25	may be solved by considering other related compounds,
00:05:31.10	and will reverse course and move back out into the clinic.
00:05:35.11	But many compounds -- very many compounds --
00:05:37.10	have in fact fallen out of the pipeline altogether.
00:05:43.07	And that emphasizes the need
00:05:45.16	to identify new compounds to feed this pipeline
00:05:48.07	if we're going to have any compounds
00:05:50.14	coming out of...
00:05:52.05	out into the clinic in the years to come.
00:05:58.05	And the identification of drug targets
00:06:00.26	poses some severe challenges.
00:06:03.29	So, let's consider, for example, the identification...
00:06:07.13	one strategy for identifying targets,
00:06:09.05	what the information is that one might want to know.
00:06:12.29	Let's consider the various attributes
00:06:17.09	that would be of interest.
00:06:18.24	We can certainly imagine that we would be looking
00:06:21.00	for enzymes that will be annotated from the...
00:06:22.26	from the genome sequences for organisms of interest;
00:06:27.16	various features that we've looked at, for example,
00:06:29.23	whether proteins are secreted or how large they are
00:06:33.17	or whether they're predicted to be soluble.
00:06:36.07	We could take advantage of functional genomic approaches
00:06:38.24	to ask whether the candidate targets
00:06:41.24	for drug development
00:06:43.28	are expressed at the right time or in the right place,
00:06:46.19	and we might be interested in structural information
00:06:50.06	on which we could consider modeling
00:06:54.08	particular drugs that might fit into active site pockets,
00:06:58.00	a promising approach in the...
00:07:00.01	in the pharmaceutical sector.
00:07:02.04	Some of the most interesting information,
00:07:04.01	the most valuable information for identifying candidate targets
00:07:06.18	for drug development,
00:07:08.21	lie in the realm of identifying drugability
00:07:11.00	or essentiality criteria.
00:07:14.02	By drugability, we mean whether it's going to be possible
00:07:16.28	to develop a drug for a particular target,
00:07:20.04	whether it has the right attributes,
00:07:23.06	whether its active site is likely
00:07:26.01	to be amenable to inhibition
00:07:28.18	with small molecule antibiotics,
00:07:30.21	whether the target, when it's inhibited,
00:07:33.03	will in fact kill the organism.
00:07:36.16	And I'm sad to say much of this information on essentiality,
00:07:38.25	on drugability, on structural information
00:07:43.00	-- much of the information we would most like
00:07:45.02	to know about candidate targets for antimalarial development --
00:07:47.08	is simply missing.
00:07:48.23	If we look, for example, for a particular enzyme in humans,
00:07:54.03	we have available information
00:07:57.06	from the annotation of the human genome,
00:08:00.02	we have information that we can calculate
00:08:02.02	automatically on the size of the...
00:08:04.29	of the protein, an enzyme in this case;
00:08:07.29	information on functional genomics data,
00:08:10.11	of whether it is expressed
00:08:12.11	and when it's expressed;
00:08:14.02	and even information on its drugability
00:08:16.13	or, in many cases, essentiality.
00:08:18.26	If we turn to neglected diseases,
00:08:22.09	such as malaria or African sleeping sickness,
00:08:25.22	caused by the parasite Trypanosoma brucei,
00:08:28.26	well, we are now fortunate in having
00:08:32.05	the complete genome sequence available
00:08:34.09	for trypanosomes,
00:08:37.01	and therefore can use the annotation
00:08:39.05	to identify the putative farnesyl pyrophosphate synthase
00:08:45.08	for this... for this organism.
00:08:47.08	And we can calculate information,
00:08:49.21	such as the predicted molecular weight.
00:08:52.17	We don't necessarily have the functional genomics data
00:08:56.29	that would tell us whether this enzyme
00:08:59.01	is expressed at the right time
00:09:01.00	and in the... in the right place,
00:09:03.17	and neither do we have structural information,
00:09:05.26	or information on drugability or essentiality.
00:09:09.24	We can, however, consider taking advantage
00:09:12.04	of the evolutionary relatedness
00:09:14.25	between these proteins in inferring orthology.
00:09:19.11	And by orthology,
00:09:21.28	what we mean is the evolutionary relatedness of proteins,
00:09:25.07	proteins that have a shared evolutionary history
00:09:29.06	and a shared function.
00:09:31.02	And if we could define a true ortholog
00:09:36.28	-- a protein with likely similar function --
00:09:40.14	then we could infer these various attributes of interest,
00:09:44.28	by identifying an orthologous group
00:09:47.17	and inferring information, for example, on drugability,
00:09:50.29	by analogy with known drug ability criteria in humans
00:09:56.04	or in yeast or in other organisms of interest.
00:10:01.13	And so, my laboratory group and many others
00:10:06.07	have invested a great deal of effort
00:10:10.10	in attempts to try to define orthologs across species boundaries,
00:10:13.23	and I'd like to briefly introduce you to another database,
00:10:17.08	and we will do this simply by virtue of screen dumps
00:10:20.06	rather than taking the time to walk through this live,
00:10:22.25	but I encourage all of you
00:10:25.23	to take a look at the OrthoMCL database
00:10:28.01	at orthomcl.org,
00:10:30.29	accessible via the Plasmodium genome database
00:10:35.17	or by... as a direct URL, here.
00:10:39.09	This database is designed to facilitate queries
00:10:42.24	across all of life, focusing particularly on the eukaryo...
00:10:47.01	eukaryotic species,
00:10:49.16	on all eukaryotic species for which complete genome sequence
00:10:52.04	is available,
00:10:53.29	and a representative selection of bacteria
00:10:56.24	and archaebacterial taxa.
00:10:59.02	So, imagine, for example,
00:11:01.15	that we were interested in ask...
00:11:03.22	in formulating an evolutionary question
00:11:06.10	related to the origin of the apicoplast,
00:11:08.04	that organelle that we described earlier.
00:11:11.00	We might in this case be interested in asking
00:11:15.21	about proteins that are derived
00:11:19.08	from a plant or algal ancestor,
00:11:23.24	present in apicomplexan parasites,
00:11:26.15	including Plasmodium parasites and other haemosporida,
00:11:31.14	Toxoplasma parasites, Theileria parasites,
00:11:35.07	but absent from Cryptosporidium,
00:11:37.12	which interestingly has lost this organelle,
00:11:41.28	but also proteins which are lacking...
00:11:45.05	have been... that are absent from any mammalian
00:11:51.12	or animal or fungal taxa
00:11:53.20	where the apicoplast is not present.
00:11:57.01	And when we run a query like that,
00:11:59.15	we come up with a list of groups of proteins,
00:12:01.19	proteins indicated here...
00:12:03.27	a few dozen protein groups
00:12:07.07	represented by the phylogenetic spectrum
00:12:09.26	across various organisms the...
00:12:12.26	from the bacteria and archaebacteria,
00:12:16.03	metazoan and protozoan organisms.
00:12:18.09	This gives us a list of several genes
00:12:21.21	including targets of fluoroquinolone antibiotics;
00:12:27.00	enzymes that are known as...
00:12:28.25	such as this DNA gyrase, two subunits indicated here;
00:12:32.18	proteins that are known to be associated with the...
00:12:35.15	with the apicoplast,
00:12:37.15	this LytB protein, for example,
00:12:39.07	a rapid means of identifying,
00:12:42.05	purely on the basis of evolutionary conservation,
00:12:44.22	the proteins that we can find
00:12:49.07	and that might be of interest, for example,
00:12:51.00	in defining metabolic pathways
00:12:53.25	associated with the apicoplast.
00:12:55.24	Turning to the question of drug target discovery,
00:12:58.14	we can use this information
00:13:01.07	-- inference based on orthology --
00:13:04.09	to fill in many of the gaps in our knowledge,
00:13:08.08	and this has explicitly driven yet another database
00:13:13.08	that I'd like to tell you about today,
00:13:16.11	the last of the databases we'll be discussing today,
00:13:18.29	the TDR Targets database
00:13:22.02	available at TDRtargets.org,
00:13:24.29	which differs from the databases that we've described thus far
00:13:28.24	in considering a much wider range of species,
00:13:30.21	organisms of particular interest
00:13:35.21	to the Tropical Disease Research unit
00:13:37.25	of the World Health Organization,
00:13:40.02	and where we can infer absent information based on...
00:13:44.19	based on orthology.
00:13:46.18	This is a project that has involved
00:13:48.13	a collaboration between many different groups
00:13:50.05	and many different funding agencies,
00:13:51.28	and I'd like to briefly take you through
00:13:54.03	a tour of some of the kinds of questions
00:13:56.21	that we can ask.
00:13:58.10	If we were to turn to the OrthoMCL...
00:14:00.21	if you were to turn to the TDR Targets database,
00:14:03.05	you would be presented
00:14:06.22	with a series of quite simple forums
00:14:09.04	focused explicitly on drug target identification.
00:14:12.17	You might, for example,
00:14:14.15	want to define targets based on predicted enzymes --
00:14:18.21	whether they have an EC number,
00:14:20.20	whether they've been annotated as being...
00:14:22.19	as having catalytic activity.
00:14:24.15	You might look for small and soluble molecules,
00:14:26.14	and I have formulated this query
00:14:28.20	to simply ask for proteins under
00:14:31.10	a predicted molecular weight of 100,000
00:14:33.03	and with no transmembrane domains,
00:14:35.02	for which a crystal structure is available,
00:14:36.29	or at least a structural model,
00:14:39.28	which are shared across related organisms,
00:14:44.11	nd in this... and in this question,
00:14:46.29	for targets in the Trypanosoma brucei parasites
00:14:53.04	responsible for sleeping sickness,
00:14:56.16	I've asked for proteins that are found in related parasites,
00:15:00.27	Leishmania major, responsible for kala-azar,
00:15:02.25	Trypanosoma cruzi, responsible for Chagas disease,
00:15:06.03	as well as Trypanosoma brucei,
00:15:08.28	but absent from humans,
00:15:10.14	and one might imagine absent from some of these other organisms
00:15:13.09	as well...
00:15:15.17	looking for proteins for which we have,
00:15:18.06	if not essentiality data in trypanosomes,
00:15:22.09	where we have very limited essentiality data,
00:15:24.21	any evidence of essentiality in any other organism
00:15:28.10	-- in yeast, in E. coli, in other bacteria, in mice --
00:15:36.15	evidence of essentiality
00:15:39.13	based on a variety of tools
00:15:42.05	developed in the pharmaceutical industry
00:15:45.14	for drugability or for the nature of the chemical compounds
00:15:49.01	which target, in this case,
00:15:51.06	orthologs of these proteins,
00:15:53.05	where there is curated information
00:15:55.16	validating the phenotype that's observed when this...
00:16:04.19	when this enzyme is targeted,
00:16:06.18	whether we have literature references or assays,
00:16:10.00	for example, that are available for evaluating performance.
00:16:14.11	And if we turn back to the page
00:16:16.14	and ask this long series of questions...
00:16:19.04	we can once again look at the history of the questions
00:16:22.02	that we've asked,
00:16:23.19	just as we did in considering the Plasmodium genome database...
00:16:26.18	only in this case, in this list of queries
00:16:29.17	we've also... we've also introduced the ability
00:16:33.11	to weight those queries.
00:16:36.06	So, in this particular query,
00:16:39.07	I've insisted on proteins being enzymes --
00:16:42.02	I've given that a very high weight.
00:16:45.03	I've also placed very high value on targets
00:16:50.19	that have been manually curated
00:16:53.14	and those for which we have crystal structure information --
00:16:57.13	structural models are potentially valuable,
00:16:59.19	although perhaps not as useful as a crystal structure --
00:17:02.15	and a whole series of other weights.
00:17:05.07	And the net result of this is the ability
00:17:07.15	to produce a long list of prioritized targets
00:17:12.23	with weights that are applied from the most...
00:17:16.26	from the most promising
00:17:20.00	down to the least promising.
00:17:21.13	And we can consider this long list of targets
00:17:27.00	as a ranked and weighted list,
00:17:28.18	returning as desired to the...
00:17:31.24	to the weights that we've applied,
00:17:33.13	modifying those, modifying our queries.
00:17:35.17	And finally, we can take these lists
00:17:37.12	and post them for others
00:17:39.24	who may want to explore the list
00:17:42.18	or modify the various current of criteria
00:17:45.20	that one would be interested in looking at.
00:17:48.09	And in fact, there are many posted lists of candidate targets
00:17:51.11	that you may wish to explore
00:17:53.07	if you're interested in targets for trypanosomes,
00:17:56.09	for malaria,
00:17:57.28	or for various other organisms.
00:18:00.10	Now, I'd like to close with an interesting story that highlighted,
00:18:06.03	at least for me, the potential of these sorts of approaches.
00:18:10.02	Recently I was giving a presentation
00:18:12.27	on this TDR Targets database,
00:18:15.18	and I was discussing candidate targets for...
00:18:20.01	candidate targets for drug development
00:18:22.13	and formulated a query a little bit more extensive
00:18:25.04	than the queries we've looked at before.
00:18:27.15	This is a query with 14 parameters
00:18:30.12	and a variety of associated weight criteria,
00:18:34.18	giving rise to a long list of candidate targets
00:18:39.14	for trypan... antitrypanosome drug development,
00:18:43.26	including many targets that are the subject of intense exploration
00:18:51.29	as potential new drug targets for African sleeping sickness.
00:18:55.24	And a hand went up in the audience
00:18:57.19	from a colleague who said, you know,
00:18:59.05	this looks very nice, but to be honest,
00:19:02.15	most of us in the audience are...
00:19:06.19	work on other organisms,
00:19:08.09	on malaria or on tuberculosis.
00:19:12.00	What would happen if...
00:19:14.24	how would we go about exploring potential targets there?
00:19:18.05	And without knowing the answer to that question,
00:19:21.15	I said, let's take the same set of criteria
00:19:25.16	about crystallographic information
00:19:28.24	and annotation and solubility and drugability
00:19:32.29	and simply -- blind -- apply the same criteria
00:19:36.00	to Plasmodium parasites
00:19:37.22	and to the bacteria responsible for tuberculosis.
00:19:42.24	And it was certainly gratifying
00:19:45.18	to find that top on the list of candidate mycobacterial targets
00:19:48.20	was INHA, the target for isoniazid.
00:19:52.12	Second on that list
00:19:54.21	was dihydrofolate reductase.
00:19:55.28	Third on that list was one of the enzymes
00:19:58.25	associated with lipid metabolism.
00:20:01.06	In fact, it's the same pathway
00:20:03.17	associated with the apicoplast in Plasmodium
00:20:05.25	and a promising target for drug development.
00:20:07.24	So, this served to nicely validate
00:20:10.13	the promise of these kinds of strategies,
00:20:13.15	and I hope that each of you will consider
00:20:16.08	exploring genome databases
00:20:19.16	as ways of carrying out computational questions,
00:20:28.25	computational experiments,
00:20:31.12	to formulate hypotheses that can be tested further
00:20:34.14	in the laboratory
00:20:36.02	and bringing to bear the wide range of techniques
00:20:38.11	that we have available.
00:20:40.10	Now, needless to say, these resources,
00:20:46.21	and we've talked about...
00:20:48.28	we have talked about several different databases
00:20:51.17	over the course of...
00:20:53.11	over the course of this lecture...
00:20:54.28	the PlasmoDB database that's a component of the apicomplexan database
00:20:58.05	and the EuPathDB database,
00:21:00.27	the ortholog database
00:21:04.14	that's used for interrogating across all of life
00:21:07.13	and the TDR Targets database
00:21:09.12	focused on drug target development...
00:21:12.08	all of these resources
00:21:16.20	rely on input from many investigators in my laboratory
00:21:18.29	and elsewhere.
00:21:21.24	and I certainly won't take the time to read
00:21:23.12	the long list of names that's involved.
00:21:25.24	But most importantly, I'd like to acknowledge,
00:21:27.12	from the standpoint of Plasmodium,
00:21:29.18	the Plasmodium genome consortium
00:21:33.29	and, from the standpoint of all of these databases for all organisms,
00:21:37.12	that the many researchers worldwide
00:21:41.08	who've been responsible for generating and making available
00:21:44.01	the data on which these...
00:21:46.17	on which these tools are based.
00:21:51.10	And so, as we come to the end of this third lecture
00:21:54.21	on apicomplexan parasites
00:21:56.29	-- their biology, their evolution,
00:22:00.22	and the exciting new opportunities
00:22:04.01	for research in parasitology
00:22:07.10	and many other... in many other areas --
00:22:11.04	I'd like to end on a final note
00:22:14.04	that is particularly relevant to researchers
00:22:17.07	in countries that suffer from these parasitic diseases,
00:22:22.22	pointing out that particularly with respect
00:22:25.08	to the genome database project we've discussed,
00:22:30.21	there are tremendous opportunities
00:22:32.29	for working anywhere in the world,
00:22:35.05	even in environments that may be...
00:22:36.29	with... have limited infrastructure.
00:22:40.11	The computational approaches that we've described,
00:22:43.20	I hope you will agree,
00:22:45.21	offer fantastic opportunities for biological
00:22:49.14	and therapeutic perspectives
00:22:51.07	on all organisms, pathogen or not,
00:22:54.20	and capitalize on computational tools
00:23:02.03	that are universally available throughout the world.
00:23:06.25	Unlike some of the sophisticated molecular genetics
00:23:11.14	or cell biological approaches
00:23:13.08	that we've discussed
00:23:15.09	that rely on very expensive equipment
00:23:16.25	that's of limited availability,
00:23:19.04	computers are inexpensive and are ubiquitous,
00:23:22.03	and can be used even in areas
00:23:26.07	that have no reliable power
00:23:29.20	and only occasional low-speed internet access.
00:23:35.01	Moreover, as a relatively new discipline,
00:23:37.11	the area of computational biology
00:23:40.04	is certainly ripe for exploitation,
00:23:43.01	and the demand for researchers
00:23:45.07	conducting such work is certainly likely to remain high.
00:23:50.09	And because most of the communication
00:23:53.26	in this area is net-based,
00:23:55.26	it diminishes somewhat
00:23:59.02	the critical mass issues
00:24:01.22	that pose a challenge for researchers
00:24:04.19	trying to work in environments
00:24:07.04	without a rich and diverse research infrastructure.
00:24:11.29	As an area that overall is capital intens...
00:24:15.01	is intellect intensive rather than capital intensive
00:24:17.22	or technology intensive,
00:24:19.09	I think computational biology
00:24:21.26	is particularly exciting for work
00:24:24.06	wherever you may happen to live,
00:24:26.00	in areas where malaria or trypanosomes
00:24:29.00	are endemic,
00:24:31.08	in areas... as well as in areas
00:24:34.27	with a rich source of technological and academic infrastructure.
00:24:40.26	So, I hope you've enjoyed this series of lectures
00:24:43.03	as part of the iBio seminar series,
00:24:45.28	and that you will explore other iBio lectures as well
00:24:50.14	and consider ways in which the kinds of approaches
00:24:53.07	that we've talked about today
00:24:55.11	might be applied to the experimental systems
00:25:00.11	that you find of interest,
00:25:02.19	and perhaps consider ways in which
00:25:05.18	the approaches that you have taken yourself,
00:25:07.19	in your own work,
00:25:09.07	or that you've learned about in other iBio lectures or elsewhere
00:25:13.18	can be applied to the various systems that I've talked about today.
00:25:15.16	And I look forward to hearing more about the work
00:25:17.26	that you do and to following the develop...
00:25:23.16	the continued development of this area in a time
00:25:27.25	where the rapidly advancing technology
00:25:31.07	in cell biology, in molecular biology,
00:25:33.15	in computational biology
00:25:35.14	has offered many opportunities for all of us.
00:25:37.26	Thanks very much for your attention,
00:25:39.14	and best wishes with your further studies and research.