Bioinformatics

Dr. Keenan Amundsen Author

01/07/2016 Added

179 Plays

Description

Bioinformatics presented by Dr. Keenan Amundsen

Searchable Transcript

Search:

[00:00:01.073]Today I'm gonna frame my talk
[00:00:03.778]using buffalograss as a model,
[00:00:06.077]and so that's the primary species that
[00:00:07.759]our lab works with, so basically I'm gonna
[00:00:10.372]chat a little about that.
[00:00:11.533]Before I do, though, I think it's important
[00:00:14.028]to revisit why buffalo,
[00:00:16.420]or bio-informatics, there's too many "b" words in this talk,
[00:00:19.218]so I'm gonna get confused.
[00:00:20.785]But why bioinformatics is kind of emerging
[00:00:23.606]as an important field.
[00:00:25.487]And I'm gonna do that by looking at some
[00:00:27.808]publicly available data.
[00:00:29.666]I pulled some information that's available
[00:00:33.427]from the National Center of Biotechnology Information,
[00:00:36.225]which is a public repository.
[00:00:37.827]Stores a lot of DNA sequence information,
[00:00:40.660]along with lots of other information,
[00:00:42.540]but for this talk right now I'm just gonna focus on
[00:00:45.512]some of the expressed sequence tagged information
[00:00:48.447]that they have.
[00:00:49.795]So basically, what I'm showing here
[00:00:51.433]is the amount of expressed sequences
[00:00:55.113]that have been sequenced
[00:00:56.541]and deposited in the database for wheat.
[00:00:59.640]And so you can see that there's about
[00:01:01.103]one and a half million, I think that's right.
[00:01:03.541]Yeah, so about one and a half million
[00:01:05.155]EST sequences for wheat.
[00:01:07.012]That's a lot of data, right?
[00:01:08.823]And so that's part of what we're talking about
[00:01:10.545]for bioinformatics.
[00:01:11.657]It's a recurrent theme that
[00:01:12.875]came up with some of the talks this morning.
[00:01:14.802]It's really a lot of data.
[00:01:16.938]And also, to look at some of these other species
[00:01:19.956]like soybean, it's about the same.
[00:01:22.174]As for wheat, and then corn's up over two million.
[00:01:25.540]But I don't have the luxury of working with those species,
[00:01:28.814]unfortunately, so I work primarily on buffalograss.
[00:01:31.808]As of last week when I pulled this data
[00:01:34.885]there's zero ESTs published for buffalograss.
[00:01:39.529]What I'm gonna do now is talk about some of the
[00:01:41.579]challenges of using some of these tools
[00:01:44.950]to explore, basically specialty crops
[00:01:48.218]that don't have a lot of the same types
[00:01:49.756]of resources as some of these bigger agronomic crops.
[00:01:54.446]But again, to do that I think it's nice to
[00:01:55.888]set the stage and give everybody a little background
[00:01:58.871]on where, or how some of these tools have evolved.
[00:02:02.563]And I know some of that discussion came out
[00:02:04.246]this morning.
[00:02:06.110]But this graph here I completely stole from the internet,
[00:02:10.284]but basically it shows growth in
[00:02:16.764]sequencing technology over the past 30 or so years.
[00:02:20.458]And you'll notice,
[00:02:22.258]on the horizontal axis this is showing years
[00:02:24.719]since 1980 to the future, a future date.
[00:02:28.097]And the vertical axis there is showing
[00:02:32.137]the amount of sequence data that one of these
[00:02:35.584]sequencers can generate, and so we're talking
[00:02:37.918]really about analyzing and working with large
[00:02:41.423]scale sequencing data sets.
[00:02:43.583]And the other thing that it's kind of
[00:02:45.557]important to point out is
[00:02:47.321]you'll notice that the graph on the,
[00:02:50.398]or the vertical axis is basically on a logarithmic scale.
[00:02:53.857]So we've seen this huge growth
[00:02:55.842]over recent years in particular,
[00:02:57.340]with the amount of data that these
[00:02:59.800]sequencers can generate.
[00:03:01.333]So that's an important concept that we'll talk about.
[00:03:05.242]And then, they're developing these sequencers
[00:03:08.055]basically as fast as they can put one of these
[00:03:10.334]graphs together,
[00:03:11.375]so they're really coming in the market really quick.
[00:03:13.662]One of my favorites is this little guy here.
[00:03:15.938]It's called the MinION.
[00:03:18.143]And it's no bigger than my cellphone,
[00:03:20.768]but it has the ability to generate tons and tons of data
[00:03:24.191]and it's basically just a USB key that you plug in,
[00:03:26.722]you add your sample to it.
[00:03:28.104]You're able to generate some data.
[00:03:29.903]And since my kids are kind of like-minded,
[00:03:32.504]they also do a lot of Google searching for
[00:03:34.790]MinION, but of course they pronounce it
[00:03:37.043]a little differently.
[00:03:41.310]When we're talking about Big Data, I know Harkamal
[00:03:43.230]touched on it this morning, as did Daniel
[00:03:45.413]and some other folks, right.
[00:03:46.978]This is just an example of some of the
[00:03:49.453]sequence data that our lab generated.
[00:03:51.357]Just to kinda give you an idea of how much
[00:03:53.575]data we're really talkin' about.
[00:03:55.683]We work with these horrible, non-graphical
[00:03:59.053]terminal emulators when we're analyzing this data.
[00:04:02.014]And so, on this particular slide here,
[00:04:05.079]that screen capture is just showing
[00:04:08.562]four sequences, which isn't much, right?
[00:04:11.974]But this particular project, on this particular
[00:04:16.293]data file had more than 12 million sequences,
[00:04:19.172]just in one data file, right,
[00:04:20.820]and so that'd be the equivalent of,
[00:04:22.841]I think my math's about right.
[00:04:24.013]So about three million screens of data.
[00:04:26.810]That's a lot of information,
[00:04:28.958]and it's just one text file,
[00:04:30.560]and the text file itself is several gigabytes in size.
[00:04:33.938]So, you can't even open these
[00:04:35.818]in a lot of, you know, Microsoft Word,
[00:04:38.094]or something like that.
[00:04:39.232]So there's tons and tons of data.
[00:04:42.238]And then, for this particular project
[00:04:45.837]we had 36 such data files.
[00:04:47.718]Harkamal talked about one that had, I think he said
[00:04:49.715]800 or so.
[00:04:50.609]Some of these projects get ridiculously big, and they're
[00:04:54.010]generating ridiculous amounts of data, okay.
[00:04:57.458]And so those are definitely some of the challenges.
[00:05:00.197]A lot of the early drivers of this increase in technology,
[00:05:03.913]so we're really talking about increase in technology,
[00:05:06.153]we're able to generate a lot more sequence data
[00:05:08.754]than we could five years ago,
[00:05:11.029]10 years ago.
[00:05:12.248]These are just a couple of projects that are driving
[00:05:15.011]some of this innovation.
[00:05:18.565]And honestly, there's hundreds or thousands more
[00:05:21.454]of these types of projects.
[00:05:23.079]You might imagine that we've seen
[00:05:26.341]a lot of growth in the technology to generate this data,
[00:05:29.139]but really the holdup is in bioinformatics
[00:05:32.053]and computational biology.
[00:05:33.353]There's a huge bottleneck in being able to
[00:05:36.199]analyze the data,
[00:05:37.858]and so that's becoming the problem.
[00:05:39.401]And so now we're at a place where we can
[00:05:41.178]generate as you saw on that previous slide,
[00:05:43.674]or a couple of slides ago,
[00:05:45.681]billions and billions of base pairs of sequence data
[00:05:48.781]just in one sequencing run.
[00:05:50.396]But the problem is, having the expertise,
[00:05:52.834]the bioinformatics knowledge and
[00:05:54.134]computational biology knowledge to be able to
[00:05:56.060]analyze that data, and so that's really the challenge.
[00:06:00.856]I kind of have a background in bioinformatics
[00:06:02.503]and computational biology, so that's why
[00:06:04.001]Roch asked me to talk about this today.
[00:06:09.034]I do want to make one distinction, though.
[00:06:10.723]We're kind of, I don't know that we have
[00:06:12.371]any true bioinformatics people in the department.
[00:06:15.911]So most of us that are on this side of things
[00:06:18.582]would argue that we're computational biologists.
[00:06:20.532]So, kind of a true definition
[00:06:23.023]is that bioinformatics folks are the ones that are
[00:06:26.499]developing new algorithms and software tools
[00:06:29.448]to analyze some of the data, whereas
[00:06:31.131]computational biologists are actually using the tools
[00:06:33.534]to analyze the data sets, and so
[00:06:35.450]I definitely fall more into that later category.
[00:06:39.655]As I mentioned at the beginning, one of the things
[00:06:40.980]I wanted to do is just talk about
[00:06:42.909]some of our research in buffalograss
[00:06:44.289]and how we can take advantage of some of these
[00:06:46.552]modern sequencing tools to try and study
[00:06:49.792]a plant system for which there's really no
[00:06:53.612]up front genomic information, okay,
[00:06:55.851]and so that's what I'm going to spend
[00:06:57.071]a little time on.
[00:07:00.138]And what I really want to do is just focus in on
[00:07:01.912]just one project, we have several of these going,
[00:07:04.268]but this is probly the one that's furthest along
[00:07:06.974]in the process and I think is, we've have some
[00:07:09.156]interesting data from it.
[00:07:10.747]Leaf spot disease, it's caused by a disease complex.
[00:07:14.774]I work with buffalograss, it's a low input, sustainable
[00:07:17.432]turf grass species.
[00:07:18.989]It's native to the Great Plains region.
[00:07:20.904]Leaf spot disease is, again,
[00:07:23.005]caused by several pathogens,
[00:07:24.527]but it's one of the diseases that's important
[00:07:26.558]for buffalograss.
[00:07:30.215]Under the right conditions or, as a manager,
[00:07:32.663]the bad conditions,
[00:07:34.301]so under the right conditions you can get disease that's
[00:07:37.203]severe enough to cause stand loss, and so that's a problem.
[00:07:39.979]One of the things we did,
[00:07:41.859]I also am a turf grass breeder, and so
[00:07:44.575]we want to identify new sources of host resistance
[00:07:47.883]to leaf spot disease.
[00:07:49.380]And so I had a post doc, Sah-Jee Bah-Rom-Rah-Dah-Sah
[00:07:52.237]that screened 84 different buffalo grass genotypes
[00:07:56.393]in the green house, challenged 'em with
[00:07:58.760]one of the pathogens that cause leaf spot disease
[00:08:01.419]and as you can see on the right, we found
[00:08:02.800]some that showed good resistance
[00:08:05.158]to leaf spot disease in the green house
[00:08:06.794]and others that were sensitive to it.
[00:08:09.151]And so we picked some of those excessions
[00:08:11.841]that were either resistant to the disease
[00:08:14.383]or highly susceptible to the disease,
[00:08:16.867]and in fact we picked two of each
[00:08:18.620]and then we did one of these high-throughput
[00:08:20.711]sequencing experiments.
[00:08:22.150]We either grew those four plants, the two
[00:08:24.134]resistant or two susceptible genotypes,
[00:08:26.293]either under control conditions
[00:08:27.629]or challenged them with the pathogen and then we
[00:08:29.614]extracted RNA, we converted that to CDNA
[00:08:33.492]and we sequence 'em.
[00:08:34.386]And so that way we can basically look at,
[00:08:36.533]take a snapshot and look at
[00:08:37.972]all of the genes that are turned on
[00:08:41.130]in response to the pathogen
[00:08:42.674]at that one point in time.
[00:08:45.089]So that's kind of the idea.
[00:08:46.424]So what we're doing here is we're looking at,
[00:08:48.363]basically 190,000 transcripts at one shot
[00:08:52.508]and we're looking at how they differ between
[00:08:55.167]the susceptible lines and the resistant lines.
[00:08:58.346]And so one of the ways that we think is kinda creative,
[00:09:02.746]one of the ways that we looked at this, again,
[00:09:05.091]buffalograss doesn't have any up front information
[00:09:07.275]and so, it limits what we can do a little bit with it.
[00:09:10.756]So we turned to a good friend, foxtail millet,
[00:09:13.855]and we mapped all the differentially expressed genes
[00:09:16.271]to the foxtail millet genome.
[00:09:18.674]And I think in this graph here the little red bars
[00:09:22.331]and the blue bars show
[00:09:25.409]where those differentially expressed genes
[00:09:26.913]map to foxtail millet, and this shows the
[00:09:29.399]nine chromosomes of foxtail millet there.
[00:09:32.949]The length of the bars represent
[00:09:34.817]the magnitude of change in gene expression
[00:09:37.405]when we compare the control to the inoculated conditions.
[00:09:40.691]And, admittedly this is a busy graph
[00:09:43.012]and I don't expect you to take anything from this,
[00:09:44.847]but basically we had a good distribution
[00:09:48.029]of differentially expressed genes throughout
[00:09:50.066]the foxtail millet genome.
[00:09:52.869]And so if we just take a closer look
[00:09:54.023]and look at chromosome eight, you might notice in that
[00:09:57.021]six o'clock position there,
[00:10:00.359]there are a handful of genes
[00:10:01.991]that are turned on
[00:10:04.022]in the two resistant lines shown in the red,
[00:10:06.761]and then there are no genes in the same
[00:10:08.700]position that are differentially expressed
[00:10:10.511]in the susceptible lines, right, that are shown in the blue.
[00:10:15.401]And so, to us that suggests that that's potentially
[00:10:17.350]a genomic region of importance.
[00:10:19.032]So even though we're using a Setaria italica,
[00:10:22.573]this foxtail millet as a reference,
[00:10:24.801]not buffalo grass, it gives us some insights
[00:10:26.927]on genomic regions of potential importance.
[00:10:29.644]So that's one of the ways that we're using
[00:10:31.511]some of this data.
[00:10:32.963]Another way is to do a traditional differential
[00:10:35.529]gene expression type study, and so again I mentioned
[00:10:38.001]that there are a hundred and I don't remember
[00:10:39.649]how many thousand genes that we're looking at
[00:10:43.005]under these conditions.
[00:10:44.770]And in this graph we're looking at
[00:10:46.744]the changes in expression, or comparing
[00:10:49.425]the resistant lines to the susceptible lines.
[00:10:52.407]So, Prestige on the bottom is
[00:10:56.522]sensitive to leaf spot disease, so that's the
[00:10:58.620]horizontal axis, anything on the right
[00:11:00.337]is considered Up regulated in response to the pathogen.
[00:11:03.863]And then, anything on the left is Down regulated
[00:11:06.405]in response to the pathogen.
[00:11:08.243]And then, 95-55 on the vertical axis is our
[00:11:11.865]resistant line, so anything above that midpoint line
[00:11:15.882]is being induced in response to the pathogen.
[00:11:19.527]And, so you can imagine
[00:11:25.589]a diagonal line there,
[00:11:26.968]that's essentially showing that most of the genes
[00:11:30.753]are behaving the same in both
[00:11:32.399]the resistant and susceptible lines.
[00:11:33.852]But the ones that really caught our eye
[00:11:35.489]and that were most interesting to us are these ones
[00:11:37.300]that are shown in that red circle there that were
[00:11:39.703]Up regulated in response to the pathogen
[00:11:42.164]in the resistant line, 95-55, and Down regulated
[00:11:46.007]in Prestige, and so, we developed some
[00:11:48.351]genetic markers.
[00:11:49.431]We took a closer look and looked at the expression
[00:11:51.556]of those, did kind of a typical heat map,
[00:11:54.110]and so this is representing those 536 genes
[00:11:56.954]in that little red circle and looking at
[00:11:59.230]how they're expressed.
[00:12:01.029]The red indicates that they're Down regulated,
[00:12:02.874]green indicates that they're Up regulated.
[00:12:05.172]And you can see that a lot of these genes
[00:12:06.705]are Up regulated in the resistant lines,
[00:12:08.597]Down regulated in the sensitive lines.
[00:12:10.536]So that's kind of what we saw.
[00:12:12.297]Doing these kinds of studies
[00:12:14.157]and taking this kind of approach,
[00:12:15.435]it gives us a lot of information
[00:12:17.048]in buffalograss, and again we had
[00:12:19.881]no up front information on buffalograss,
[00:12:22.226]so in our eyes, using some of these modern
[00:12:25.627]sequencing tools, it's pretty powerful.
[00:12:27.845]So, again, can gain a lot of information.
[00:12:32.829]Just kind of quickly here, we have several other
[00:12:34.903]projects going on, looking at different traits
[00:12:37.666]in buffalograss, we're looking at gender expression,
[00:12:39.721]buffalograss is a diecious species,
[00:12:41.578]so we're looking at
[00:12:43.145]comparing male buffalograss to female buffalograss
[00:12:45.700]and finding genes important for
[00:12:47.942]gender expression.
[00:12:49.242]We're looking at leaf spot disease resistance,
[00:12:51.586]like I just showed, chinch bug resistance.
[00:12:53.620]Just initiating a new project now where we're
[00:12:56.184]looking at mechanisms of seed dormancy
[00:12:58.586]and using some of these technologies to do that.
[00:13:02.035]And then, again, I'm managing a buffalograss
[00:13:04.972]breeding program, and so
[00:13:07.177]my primary interest is in using all these tools
[00:13:10.556]to develop genetic markers or learn some type
[00:13:14.259]of information that we can apply to
[00:13:16.569]improve the efficiency of our breeding program
[00:13:18.613]to select, basically, either germplasm
[00:13:22.931]that has new sources of resistance
[00:13:25.217]or develop markers that we can track
[00:13:27.969]resistance through our breeding population.
[00:13:29.792]So that's kind of the idea.
[00:13:31.196]Here's an example where we found some markers
[00:13:33.368]that could do that.
[00:13:34.633]These are gene expression based markers
[00:13:36.189]that we used.
[00:13:37.895]And again, the idea is then to be able to track those
[00:13:41.134]and ultimately develop new buffalo grasses.
[00:13:43.316]That's kind of the idea.
[00:13:44.872]I promised Roch I would talk quick.
[00:13:48.147]Try and get us back on track,
[00:13:49.017]so that's all I have, so thanks.
[00:13:50.294](applause)

The screen size you are trying to search captions on is too small!

You can always jump over to MediaHub and check it out there.

Comments

0 Comments

Bioinformatics

Description

Searchable Transcript

Comments icon comment

Related Channels

Comments