0:01
Hi, good afternoon.
0:02
Welcome to this section.
0:05
It's Advanced Sequencing Technology.
0:08
My name is Yanping Zhang.
0:09
I'm a Core Director at the UF ICBR Gene Expression & Genotyping.
0:14
I will be chairing this section, and our first speaker is Doctor Elizabeth Louie. Dr Elizabeth Louie
0:23
earned her PhD in Molecular and Cellular Pharmacology from Stony Brook University and then joined GENEWIZ in 2011.
0:33
Doctor Louie provided consultative and the project design support for researchers and the investigators focus on genomics, transcriptomics and proteomics solutions.
0:55
All right, everyone, thank you for your time.
0:57
Hopefully no one will fall asleep after lunch.
1:00
So I'll try to keep this as engaging as possible.
1:03
But yeah, so I am from Azenta or some of you may know us as a GENEWIZ.
1:09
And today we're going to be talking about generally a case study that we have, regarding utilising a single sample and the ways that you can try to figure out how to use your samples that might be precious in the amount of material that you have to be able to run an actual multi-omic analysis.
1:27
So before I actually jump into the kind of the assays and potential assays that are out there, I just want to talk about biomarkers in general.
1:35
Of course, you know, why are biomarkers so important and why are we studying them?
1:40
Well, one, it's, you know, it's a great way to be able to measure what something is, how something is happening within your specific sample.
1:51
So being able to phenotype it, being able to determine what is actually causing your disease or whatever condition that you're actually seeing.
2:04
So a lot of the work that I'll be talking about today is going to fall under the category of molecular biomarkers.
2:10
So that's really going to involve genetic screening, epigenetic screening, transcriptomics, proteomics and metabolomics type of screenings.
2:22
So, and when you think about the central dogma of genetics and how things work, it's really important to think about all of the different stages that you're studying.
2:32
Because when you look at genomics, you have, you know, a huge genome sequence, 20,000 or so genes of a human genome.
2:41
You know, that's, billions and billions of potential variants.
2:46
And then when you kind of look at, well, what happens with the genomics, well it's now potentially being influenced by epigenomic factors, which is, you know, a smaller subset.
2:58
of potential influences, but there's a lot of different things that could be happening.
3:04
And then you look at the transcriptomics and again you have a lot of genes that can be affecting it.
3:08
And then proteomics and at the proteomics level, you're looking at like millions of potential proteins and actual molecules that could be influencing the phenotype of that sample.
3:24
So what we really set out to do was to work with our customers to find a way to be able to assay all of these things with utilising as little material as possible.
3:37
And really the main part of this is going to begin at that study planning and that sample sourcing stage.
3:44
So I know a lot of times you have existing samples, maybe in a bio bank in archive or you're collecting fresh samples and you have a primary assay in mind.
3:56
In an ideal world, you're really thinking about all of the assays and you know, the best case scenario that you could potentially run doesn't mean that you don't have a primary assay, but it means that you're also considering - like if funds were not an issue in the future, you know, what other assays can we be running and how should I store them?
4:16
And what types of considerations of the types of assays can we be running, you know, with the limited amount of sample that we have.
4:23
So here, as I said, we're going to try to focus on ways that you can analyse the genome, the epigenome, the transcriptome and the peridium.
4:35
So first, we can focus on the genome and the epigenome, and there's a lot of different assays that are available, but the main assays that Azenta and GENEWIZ technically try to recommend is either an unbiased whole genome sequencing workflow.
4:54
This is typically what we call like the gold standard, right?
4:56
A lot of people are doing it.
4:59
There's a long read and short read depending on how your sample was stored, what conditions there are.
5:06
You know, if you have like FFPE samples and you still want to do whole genome sequencing, you can't really do long read sequencing.
5:13
So you've already, you know, kind of limited yourself to doing a short read workflow.
5:19
You know, other things to consider is how much material do you have for whole genome sequencing?
5:24
Do you actually have enough to get the coverage that you've really want?
5:27
Are there specific libraries or assays that you need to consider or limit yourself to just because you don't have the right amount of material for a specific platform?
5:38
But whole genomes are really powerful in this case, just because, as I said, it's unbiased.
5:44
You don't need to have a preconceived notion of the potential biomarkers or targets that you're hoping to look for.
5:51
On the other side though, you know, we do also do whole exome sequencing as well as a targeted workflow.
5:58
And these really shine in a few different ways.
6:01
One is you can get a lot deeper coverage because you're only targeting the coding areas of a sample versus the entire genome.
6:10
But you do need to have some sort of preconceived notion of like what you're actually looking for and what you're searching for.
6:17
But the other thing with, you know, targeted sequencing and whole XM sequencing is that you can get really high-quality data if you have like a more degraded sample just because of the nature of the library preps that are out there right now.
6:31
So if you do have samples that are limited in the amount, as well as samples that might not be the most intact, then using a targeted workflow works really nicely.
6:45
And then on the epigenomics side, of course, there's methylation sequencing to look at methylation sites and that can be at the whole gene level or the target level and very similar relates to kind of just the regular genetic sequencing.
7:00
It really just depends on the current state of your sample, you know, the quality of the sample.
7:05
And even with the methylation condition, there's like bisulfite sequencing, there's enzymatic sequencing and there's a lot of other options there to actually be able to detect methylation.
7:19
So with methylation, you know, in some cases you may want to do like native sequencing on like a PacBio workflow where you can get both the genetic sequencing and the methylation status all in a single run.
7:35
But of course for that you need to have really high-quality DNA.
7:40
If you have less high-quality DNA, then you could potentially use more of like a enzymatic methylation workflow, in which case it's a lot more lenient towards the sample compared to like a bisulfide workflow.
7:56
And then it would also be short read to kind of like just account for the fact that you have potentially a degraded or fragmented sample. And then you have a ATAC-Seq.
8:06
That's another good way to kind of look to seeing- like it doesn't quite answer the same question as methylation, but a very similar question on just being like what is actually open and accessible to be influenced in the epigenome.
8:21
And then there's always ChIP-Seq and the limitation with ChIP-Seq, of course is the actual pull down in the antibodies that would be available.
8:32
So if we think about the transcriptome on the other side, there's of course a few different options here.
8:39
You've probably seen many variations of this slide, whether it be like fruit salad,
8:45
I've seen one as like car parts and Legos in this case.
8:50
So you know, there's a bulk option, which is really good if you're just looking at an average over a snapshot of the sample that you have.
8:59
You can do it on both like a short read and a long read workflow and it works really well even and if you have degraded samples like FFPE.
9:08
And when we talk about the bulk workflow, there's a few different options here as well, again, depending on how much sample you have and the quality of that sample.
9:19
So we listed a few that we have at Azenta.
9:22
So if you kind of just are looking at your standard gene expression workflow, kind of the most traditional workflow, it's going to fall under that standard RNA-seq or the strand-specific RNA-seq that's on your right hand side of the slide.
9:37
But if you're more interested in isoform detection and potentially integrating both isoform detection and gene expression analysis, you could do a long read sequencing like the PacBio Iso-seq, You know, coupled with their newest technology of doing like the library prep with the connects workflow to be able to get enough data to actually pull gene expression data from the original libraries.
10:05
And then if you don't have a lot of sample at all, you know, with short read workflows, there are options to use like an ultra low RNA-seq prep where you can essentially assay as little as 50 cells and still get really high-quality data.
10:20
And then if worse comes to worse, you have really poor-quality sample like FFPE samples, then we do recommend again like a targeted approach, like an RNA exome, potentially even a microarray workflow that's a little bit more targeted.
10:38
It might be not as unbiased, since you are kind of doing a more targeted detection, but you will get a lot of great useful data from it nonetheless.
10:51
And I also always include like small RNA-seq because depending on how you're actually preparing your samples, you need to think about that extraction step.
11:00
Do I actually have small RNA?
11:01
How is it influencing the rest of the transcriptome?
11:06
And then when we think about single cell, obviously there's a lot of different work flows out there.
11:14
And the workflow that we use at Azenta is the 10X workflow.
11:19
And for that, you know, we typically receive like reserve cells, but there are of course options to do more of like a fixed workflow.
11:30
So depending on the way that you've collected your samples, how you've actually had your site collected or their team collected, you can choose a specific option.
11:42
Now with the fixed workflow, what the main differences between, you know, the fixed and the fresh is the fact that it is going to be pro based.
11:49
So that does mean you are limited to whatever panels are there or you'd have to custom build a panel to potentially get your maybe like non typical species if that's what your experiment entails.
12:04
But the fixed workflow gives you a lot of flexibility because often times your samples might not survive, you know, from getting from a collection site to wherever you are processing those single cell samples.
12:25
And then you know, of course, with the single cell workflow, there's always an option to add on like immunoprofiling to actually supplement and add on additional data.
12:37
So when it comes to space spatial profiling, you know, accordingly, we typically recommend using the 10X Visium just because of the flexibility and ease of use.
12:52
And in our experience, we typically recommend doing an FFPE workflow, which again, is pro based versus fresh, but only for the consistency.
13:03
When we worked with a lot of customers, sometimes they do have fresh samples, but the issue is sometimes it's stored in different ways depending on how their collection site works or depending on, you know, how animals are sacrificed, it may not always provide the same reliable profile.
13:20
So you do have to be very careful in that case.
13:23
We find that if you to provide a fixation protocol or utilise fixation workflow, the results at the end can be a little bit more consistent from, you know, different scientists, different researchers that are doing the same work.
13:38
And then when it comes to proteomics, at Azenta we use the Olink assays now.
13:44
We chose Olink because of it's highly sensitive, it's highly specific.
13:51
It's really easy to assay a lot of sample all at once and you don't need a lot of material.
14:00
So a lot of the Olink assays, for the target assays
14:05
you probably only need like one or two microliters of whatever biofluid you're working with.
14:10
So it's really nice as an add on experiment if you've done either like a blood collection or if you have plasma that you were going to use in another experiment and you just want to add on an additional assay of Olink to kind of supplement some of the other data.
14:25
It's quite easy to do.
14:29
OK, so now I want to actually quickly go through a case study.
14:34
So I'm not going to go into the data into too much detail.
14:37
But if you are interested in seeing some of the specifics of the data, I am happy to chat with you afterwards and even send you the poster and the case study sheet that we have for it.
14:48
But essentially for this case study, what we did was we took a single blood draw from 10 donors.
14:56
And of the 10 donors, five were known to have cardiovascular disease and five were supposedly healthy.
15:03
Now, one of the complications with studying cardiovascular disease is that you actually don't know if someone has a heart condition until they something happens, right?
15:13
But they may be predisposed to it.
15:15
So, you know, you'll see that after we go through a first round of blood draws and we do some analysis, the data's a little bit, you know, iffy and we think it's because it's cardiovascular disease.
15:29
But then we do go through a second round and we get much more promising data.
15:33
And it kind of goes to show that, you know, some pre-planning in the beginning when doing the sample collection really allowed us to kind of do a second round of analysis.
15:44
But the first round what we did was we took the plasma and then we also isolated the PBMCs and with the PBMCs, we did exome sequencing, a whole genome methylation sequencing.
15:59
And then we did a single cell workflow with TCR detection.
16:04
And then we took the plasma and we did Olink profiling using a cardiovascular panel.
16:10
So just taking a quick look at some of the results that we initially got.
16:14
So what we did was we took the methylation analysis results and we compared it to the whole exome result.
16:24
And the goal here was try to kind of line up any of the potential snips and variants that we found between the two different donor categories and see if it was correlated to any hyper or hypomethylation spots.
16:41
So we did find like a few different targets, but it wasn't like as conclusive.
16:46
But we did find a few where we're like, OK, this is a potential area where we'll look into a few more papers, see what pathways are involved with to see if we can further dig deeper.
16:57
And if we, this was our lab work, we'd probably do things like knock down experiments and things like that.
17:04
We then, you know, looked at the single cell data to try to see like, OK, so are certain pathways based on those snips and those areas that, you know, are involved with those areas actually up or down regulated in certain clusters?
17:20
And also looked at the actual chain sequences to see, you know, are any of these cell types acting in a different way and are the T cells acting differently?
17:35
And then this is kind of where we paused a little bit because this is the Olink cardio metabolic panel.
17:44
And you can see that the grouping of the healthy donors are in red.
17:52
And on the left hand side, you can see like the way they're categorised and the cardiovascular disease patients are in the green.
17:59
You can tell that there's no real correlation here.
18:02
Now, we only had an N=10 and it was five and five.
18:06
And as I said, with cardiovascular disease, you can't really tell if someone's predisposed to it or if they have a genetic history of it.
18:13
And we don't have enough information on that donor profile to kind of tell how real is this?
18:20
But when we looked at it, we're like, OK, the results are super high quality.
18:23
It's really clean.
18:24
We're just not seeing what we want to see.
18:28
But this panel has only 92 proteins because it is a targeted panel.
18:33
So we decided that OK, let's take a step back, how much sample do we have left?
18:40
And let's kind of expand our assays to a much higher detection option and see what's available.
18:51
So in our second round, what we did was we added on bulk RNA-seq and then we also added on the Olink HT panel which covers about over 5000 proteins instead of the original 92 that we were kind of narrowing down on.
19:11
And then also do more proteomics using this platform or CCR analysis workflow as well as metabolomic profiling.
19:20
And what we found here was a lot more conclusive once we kind of stopped narrowing ourselves to a much more targeted assay or targeted cell type.
19:30
So we did this work in collaboration with one of our partners Panome Bio.
19:37
So the RNA-seq work was done at GENEWIZ and what we found was OK between these two donor types, 55 genes were differentially expressed between the two groups.
19:51
Then we looked separately at the SEER peptide analysis workflow and we identified 17 differentially expressed proteins.
20:04
And then we looked at the Olink HT, which again covers over 5000 proteins and biomarkers, and we found 61 differentially expressed proteins.
20:16
And then just as a comparison, if you look at the proteins between the Olink HT and the SEER mass spec workflow, there are about like 1200 proteins that have like quite a bit of overlap.
20:31
And the others are kind of a little bit more unique for the ones that were detected.
20:36
And then when we looked at the metabolomic workflow using mass spec, we found, you know, 18 differentially expressed metabolites.
20:44
So individually for all of these assays, we were able to say, OK, there's distinct differences between these two groups.
20:53
What does it really mean though?
20:56
So when you actually start to look and integrate all the data together, we find that there are 13 specific pathways that are dysregulated and seven if you just look at a few, you know, very specific assay profiles.
21:13
And with that, you know, we then kind of worked with Panome Bio to take it a step further and actually were able to find, you know, a select group of areas in the pathways that could be related to the cardiovascular disease.
21:28
And really the main takeaway here is that, you know, our first round of design was really too focused and we kind of have these preconceived notions of what we were expecting to see.
21:40
But because we actually were able to kind of prepare ourselves to be able to think of, you know, to store these samples, we were able to recover the experiment by changing our plan, but still be able to get the data that we wanted to actually find things that would be important downstream.
22:02
And with that, just to, you know, reiterate, you know, Azenta/GENEWIZ, we are a global company.
22:10
We have labs that are in the US, in Europe, in China, Japan.
22:15
And a lot of the work that I mentioned here can be done anywhere, but in the US it would be in New Jersey.
22:23
So, you know, you do have a US presence if you want to do any of these assays.
22:29
And with that, thank you for your time.
22:36
So we have time for a couple quick questions.
22:46
Thank you.
22:46
Great talk.
22:47
And I was wondering if I understood correctly.
22:50
So you mainly focus on blood samples I assume because if you're using Olink proteomics, if I'm not mistaken, it's only has been validated on serum and plasma.
23:02
So if you would use a different type of sample, you wouldn't be able to use Olink as the proteomics.
23:10
So for the Olink workflow, like the serum, plasma are the validated workflows, but there are protocols for things like live cell lysates, other bio fluids for all of those.
23:21
We actually do have some experience working with as well.
23:25
Obviously not with this case study, but you know, whatever you can think of that is a bio fluid with protein in it could theoretically be done.
23:33
OK, great.
23:33
Thank you.
23:34
Yeah.
23:36
So now I have a quick question.
23:47
And what's the minimum amount that can be used for this mode, the multi analyse?
23:54
Yeah, so that's a great question because obviously it depends on the type of assay that you have.
24:00
So we took like 10mls of blood when we did this, but we were able to get like many aliquots of plasma and of the DNA and RNA.
24:10
So with the DNA workflow, you technically need like 5 to 20 nanograms maybe and have really good results.
24:18
And then with RNA, depending on which workflow that you've chosen, right, it could be like we asked for about, you know, a million, but it could be less than cells for the single cell workflow.
24:28
And you really only need about like 20 nanograms for high quality bulk RNA-seq data.
24:32
So we had a lot of aliquots leftover just doing like the 10ml collection.
24:39
So if we wanted to actually run through more experiments, we could.
24:46
I think like for the plasma, we had like 4 to 5mls of plasma after the collection and we only used maybe like a mil of it all together for all of the assays put together, yeah.