Optimization of NGS Library Preparation: Low Inputs and Fast, Streamlined Workflows

Learn how NEBNext® Ultra is enabling library prep for multiple applications, with lower input amounts and fast, streamlined workflows. In this GenomeWeb webinar Cynthia Hendrickson from HudsonAlpha Institute of Biotechnology, Momo Vuyisich from Los Alamos National Laboratory and Daniela Munafo from New England Biolabs describe use of NEBNext Ultra in exome sequencing, bacterial genome resequencing and assembly, and strand-specific RNA-Seq.

Script

Ed Winnick:

Okay everyone, we're going to get started. Hello welcome to Genome Webinars. I'm Ed Winnick, editorial director for Genome Web Daily News and I'll be your moderator today. Today's webinar is entitled Optimization of NGS Library Preparation: Low Inputs and Fast, Streamlined Workflows. The sponsor of this webinar is New England BioLabs. Our panelists today include Cynthia Hendrickson of the HudsonAlpha institute for Biotechnology, Momo Vuyisich of the Los Alamos National Laboratory Genome Center and Daniela Munafo of New England Biolabs.

You may type in a question at any time during the webinar. You can do this through the GoToWebinar control panel, which is usually appears on the right-side of your screen. If the panel is minimized, just click on the small red box with the white arrow to expand the control panel. We will ask the panelists questions after the presentations have concluded.

We're going to start with Cynthia Hendrickson from the HudsonAlpha Institute for Biotechnology. Dr. Hendrickson, please go ahead.

Dr. Hendrickson:
Thank you. As Ed mentioned, I'm part of the Genomic Services Lab at HudsonAlpha Institute, and today I'm going to present some of our data using the NEBNext Ultra DNA sample prep kit in conjunction with exome enrichment, for simple, low-input and low-cost exome sequencing.

Next slide please.

The HudsonAlpha Institute for Biotechnology is a non-profit research center located in Huntsville, Alabama. As such the genomic services lab within HudsonAlpha is our cost-recovery next generation sequencing provider, that not only supports researchers within the institute but customers worldwide.

We offer the full-spectrum of next generation sequencing sample preparation including whole genome and target-enriched DNA library prep, as well as a variety of RNA and epigenetic sample preparations. And we offer sequencing on an Illumina HiSeq 2000, 2500 and MiSeq as well as sequencing on the Ion Torrent and 454 platforms.

Next slide please.

So one of our most popular services is exome sequencing and we typically use around 2.5 micrograms of DNA and perform a standard DNA library preparation. In this method, DNA and repair, dA-tailing, Y-Adaptor ligation and PCR amplification are performed in discrete steps each followed by a bead clean up. This method is time-consuming. It requires multiple aliquots of beads and necessitates multiple transfers of each sample which can result in sample loss and increases the number of plates or tubes required. After library preparation, we typically use NimbleGen SeqCap EZ Human Exome, however we have modified the protocol so that we can pool two libraries together per exome capture. This literally costs the exome enrichment in half. However, the method does require indexing the libraries prior to pulling for enrichment as well as using adaptor blocking oligos that are specific to these index adaptors.

However, NimbleGen does provide the sequences for index-blocking oligos specific to Illumina adaptors as part of their protcol.

Next slide please.

So instead of our using our usual DNA Library Prep, New England BioLabs invited us to try their new NEBNext Ultra DNA Library Prep method upstream of our exome enrichment. In this method, end-repair and dA-tailing occur in a single step after which a unique, cleavable hairpin adaptor which is more stable than Y-adaptors, is ligated. After ligation, a USER enzyme mix is added which cleaves a U in the hairpin generating the more traditional Y-shaped adaptors. This approach allows amplification of the library.

Next please.

Our current method takes about 3.5 hours. This new method actually removes about an hour from the process. This end prep with end repair and dA-tailing occur in one step takes 60 minutes. 15 minutes for ligation of the adaptors in which the adaptors and ligation master mix are simply added to the current reaction. After 15 minutes, USER is again simply added to the reaction and after an additional 15 minutes of incubation, there is a single bead clean up. There is the option to do a bead-based size selection in the protocol but we don't typically use size selection for exome enrichment, so we kept with a simple clean up. After that it's PCR enrichment followed by one last clean up. As a result, not only do this reduce an hour of time, but also cut the number of bead clean ups and sample transfer steps in half.

Next slide please.

To compare the methods, since we anticipated reduction and sample loss with the NEBNext Ultra, we decided to push it and instead of going in with our regular 2.5 micrograms, to go in with only 500 nanograms. But compare that to our current method with our normal inputs. Using both methods, we performed 6 cycles of PCR amplification after end repair dA-tailing and adaptor ligation. In the case of the current method, we use Kapa's HiFi Hot Start, and for the NEBNext Ultra we used their new high-fidelity master mix that's included in the kit.

QCing the results of the library we found that we got less than our inputs which we normally see and library yields with the current method was 1 microgram to 1.2 micrograms, and with the NEBNext Ultra we actually got a little bit more than we started with, with 649 and 529 nanograms. After taking some of the material for QC, we went ahead and pulled two libraries per capture: 500 nanograms each with our current method and 450 nanograms each with the NEBNext Ultra samples. The captures were performed as we normally do. We performed 15 cycles of PCR afterwards and then we found that the pool of two libraries again we had more material coming out of the NEBNext Ultra at 2.5 micrograms compared to our current method which was 1.5 micrograms.

So, going on the next slide.

For sequencing, we put them on a HiSeq 2000. We generated [inaudible 00:07:01] reads. And we'll notice that there is a difference in the number reads generated off the current method and the NEBNext Ultra. That's simply because we are a service provider so whenever we try new protocols we squeeze the samples in wherever we can. The space available for the sample in this case was a little bit less of the lane than we had for our standard current method samples.

However, aligned pairs and aligned sequence were comparable with over 99% of the reads aligning to the genome using BWA. And aligning to P19. The depth of coverage obviously reflects the number of reads generated. Quality built-in metrics were comparable and when it came to the number of duplicate reads generated, with our current method, we saw 14% and 13% which is on the high-end of our normal and with the 500 nanograms of the NEBNext Ultra samples we saw 8% and 11%, which is on the low-end of our normal.

Going on to the next slide.

So because we used 20% of our normal input, we took a closer look comparing the sequencing results and just looking at the distribution of reads across the chromosomes, you see that although of course there are a few less reads, the 500 nanogram samples, that they do basically mirror themselves in this distribution across the chromosomes. More importantly looking at the actual amount to sequence that was on target, plus or minus the fragment size, around 200 base-pairs, we see that 87% of the sequence was on target with our normal method, and 86% with the 500 nanograms which is within error or randomness. This was important for us to look at because our primary concern using a new library prep method for target enrichment is at that the adaptor blocking oligos are perfectly compatible. Seeing that the number of off-target reads were comparable between those two samples, we know that the blocking oligos were in fact equally efficient with blocking the NEB adaptors in comparison to normal Illumina adaptors.

Onto the next slide.

Taking a little deeper look, we also looked at the distribution of reads across the chromosomes. This is just randomly looking at the first arm of chromosome 1 and again we can see that the distribution of reads is comparable. Given that the 500 nanogram samples prepared with the NEBNext Ultra kit performed as well as the 2.5 micrograms prepared with our normal method, we decided to push this method further and go back to the same DNA stock. The same two individuals that we used for the 500 nanogram prep and go in with 100 nanograms just to see what would happen.

So going on to the next slide.

Comparing again, in this case it's the same individual 100 nanograms and 500 nanograms. Again you see that the distribution of reads across the chromosomes are comparable and in addition, the percent on target with again the 500 nanogram gave us 86% sequence on target. The 100 nanogram samples gave 88% sequence on target.

And then onto the next slide.

And again, zooming in and looking within the chromosome, you can see again the distributions are similar, that although the lower input, we're not seeing regions falling out or any unusual PCR bias that we wouldn't want in our samples.

So onto the next slide.

Since it is the same individual, we were able to delve deeper and we looked at the specific variants that were identified using 500 nanograms input versus 100 nanograms input. As you can see, based on the number of variants per chromosome, again the patterns were very similar. In general, the number of variants detected at the 100 nanograms were a little bit less. But, you can see that in some cases, for example chromosome 16 and 19, the number of variants in the 100 nanogram sample was a little bit higher. Specifically, a little over almost 59 000 variants were found with the 500 nanogram and 51 000 with the 100 nanogram, showing only a 13% loss in the number of variants detected with the significant reduction in DNA input.

And then onto the next slide.

And so then specifically looking at the single nucleotide polymorphisms. We found that basically we were identifying the same ones. That'll match to the db snip 135 database. You can see that over 10 000 were shared between the two samples, whereas almost equal amounts were discovered in one of the sequencing read runs versus the other. So the coverage just wasn't optimal.

And finally on the last slide.

We found that we were able to get excellent results with the NEBNext Ultra DNA sample prep kit. We were able to use DNA inputs far lower than we dared use before. We obtained greater yields post-library prep and generated some more sequencing results. In addition, the library prep was simpler and faster than our current method and required less beads and sample transfers. I would just like to just finally acknowledge everyone in Genomic Services Lab at HudsonAlpha that had some part in generating this data, in particular Nripesh Prasad who did most of the bioinformatic analysis, and also thanks to New England BioLabs for allowing us to try their new kit and share the results with you today. In particular, Pingfang Liu who gave her support in helping us incorporate this method into our exome enrichment protocol. Thanks very much.

Ed Winnick:
Okay. Thank you, Dr. Hendrickson. Just a reminder to everybody if you have a question, please type it in the question box. Should be on the right-hand side of your screen. We will pool the questions and ask the panelists at the end of the webinar. Thank you very much.

Next up is Momo Vuyisich of the Los Alamos National Laboratory Genome Center. Dr. Vuyisich, please go ahead.

Dr. Vuyisich:
Hi everyone. Go ahead and switch to the next slide. So I will be talking about our research and development projects that involved the NEBNext Ultra library prep methods that we actually tested here before they were released and we've obtained a lot of data and we're very, very happy with this library prep method and I'd like to tell you all about it. I work at the Genome Center at Los Alamos National Lab in Northern New Mexico. We have a variety of projects that are mostly R&D projects that are collaborative with universities and industries and other government agencies. You can see here the variety of projects that we do. We have a large sequencing facility that has all of the modern and GS platforms. We also have a vast computational analysis division that can handle any data for transcriptomics or metagenomics or genome assembly or analysis or anything like that.

Next slide please.

We do two types of sequencing that requires very high level of even coverage for bacterial genomes. We do genome sequencing of bacteria and here, what we do is prepare short insert shotgun libraries and we generally use Illumina for sequencing beads, and then we assemble the reads into contigs. So these are contiguous stretches of DNA that can be assembled from reads that cover the genome. You can see that in the upper left-side and on the upper right-side panel, you can see that this approach of trying to cover the entire genome is not perfect in that if you look at the y-axis, which shows the coverage of the genome, you can see that there are many gaps along the genome for which the sequence cannot be obtained using many of the library prep kits that we've used in the past. We also do genome finishing. We're well-known for that here at Los Alamos. So for that, there's a special type of library that we prepare. These are called Long Insert Paired End Libraries. They're basically 8 to 10 kb fragments that we sequence only the ends of, so we can put these contigs that we generated in the genome sequencing together into a circular chromosome that bacteria have. I will not talk about that much except that we used to use 454 sequencing for that a lot. Now we're switching to PacBio and Illumina sequencing.

Next slide please.

We had a very high-throughput bacterial genome sequencing and assembly and finishing going on until the budget cuts were introduced. Now we're just waiting for more budget approvals. This is our standard process for sequencing bacterial genomes now, using the NEBNext Ultra library prep process. We tested several commercially available kits and this one actually showed the best overall performance in pretty much every possible way so we can use very small amounts of input DNA. It is very inexpensive. It is very easy to perform. It's full-automatable and it produces data that are as good or better than any other kit available. Here on the left-side you can see the steps involved, but it's fairly simple. It's basically two consecutive reactions in the same tube followed by one purification using AmPure XP beads and then PCR amplification and PCR clean up. That's all there is to the prep basically. You can see on the right side, experimental parameters. What we did for all the kits that we tested, we applied them to three different bacteria that have three different GC contents. Silicone traces at 31, E. coli at 50 and Burkholderia strain at 67% GC content.

We also varied the input DNA amount at 10 and 100 nanograms for this kit, but we had to go up to a microgram with certain other kits. And then we also tested the ability of Betaine to improve the GC coverage during PCR. And then we varied the library fragment sizes.

If you look at the workflow, it's basically DNA quantification. We used Qubit for that, it's very accurate and sufficient.

We determined the DNA integrity of genomic DNA using E-gels, 1% EX. That's a 10-minute E-gel run. It's all pre-cast and it just takes 10 minutes.

DNA shearing, we always perform in Covaris E210. This gives us very consistent results and it's just the best process that we found.

Size selection, we performed either with AMPure XP beads or with E-gels and I will only show you the data for AMPure XP beads because they perform very well, they're fully-automatable, they're easy.

Library quantification, again we've performed with Qubit.

Library sizing we've performed with Bioanalyzer 2100, however we are now convinced that that is an unnecessary step because the size-selection is so reproducible that we can simply assume the size is always the same and simply quantify the library and generate the molar concentration of the library.

So then we performed a normalization of all libraries down to two nanomolar for Illumina sequencing. We sequence either in HiSeq or MiSeq depending on the project, and analyze the data.

So you can see on the left-side, the total amount of time is about 4.5 hours for library prep.

Next slide please.

So the first part that was important to us was to select the fragment sizes that were of interest to us. Since we normally sequence on the Illumina platform using 2 by 150 base pair sequencing. So these are paired end sequencing from both ends of the fragment, we wanted fragments that roughly are of 270 base pair lengths such that the 250 base pair reads will overlap slightly and will give us better quality of data towards the end of the reads and then we can assemble full 270 base pair fragments that are then very useful for both read mapping and assembly. You can see here that we were going for optimization of DNA sheering, to enrich for 270 to 300 base pair insert size.

So we simply dilute each DNA sample to 100 nanograms or 10 nanograms in 55 microliters and these are the cycling conditions of the Covaris E210. This is extremely similar to what NEB recommends, except that we varied the time and we observed that 100 seconds was the best for our requirements. But that's something that anyone can implement, whatever their process requires. So you can see here the size distribution of genomic bacterial DNA after 100 seconds that it's very nice and this is very reproducible. I could show you 10 of these and they would look exactly the same. That's very nice, that Covaris works reproducibly well.

Next slide please.

So you can see now, once we go through the process of NEBNext Ultra, here's one example library. We call it POA-16 for our limb system but it's Burkholderia genomic DNA, so it's two large chromosomes. And here's our prepared libraries. When we add about 270 base pair inserts to the 120 roughly base pairs of adaptors, we arrive at roughly 400 base pairs of the library fragment size average plus/minus. So this distribution is exactly what we were looking for. There are no adaptor dimers, no primer dimers. It's very clean. It sequences beautifully. Here's the summary of what we did here and what we obtained. So 100 nanograms starting material of this specific Burkholderia strain, we performed AMPure size selection, 8 PCR cycles with Betaine. Final concentration of the library is 14.9 nanograms per microliter or 56 nanomolar, so that's way, way more concentrated than necessary, so it's possible to pull back on the number of PCR cycles. Average fragment length is 429 base pairs and we sequence this on HiSeq.

You can see on the right panel, the insert-size histogram, and this is actual, based on sequencing and it's very, very accurately representing what we observed experimentally in the lab. The insert size is basically perfect for our needs.

Go ahead and switch to the next slide please.

When we look at the actual sequencing data, this is exactly the same library when sequenced, so Burkholderia bacteria have two chromosomes about 4 Mbp each. You can see here that we had low coverage overall, but the coverage is very even. It's around 50 full coverage, and you may or may not be able to see that but in the upper right-hand side of each of the coverage plots, there are some stats and they're showing that there are very few gaps and very few base pairs that are not actually mapped from the reads obtained. The first chromosome, there's only 11 base pairs out of 4 million base pairs that are not covered by the sequencing. In the second chromosome, there are 143 base pairs that are not covered. The coverage is very even, and it covers almost 100% of the genome.

Next slide please.

Here's another example, it's sample POA-15. It's the same exact strain, but we started here with 10 nanograms of the genomic DNA input. AmPure size selection again. We had to go to 12 PCR cycles and you can see the concentration of the library, is still very high, 11 nanograms per microliter, so 42 nanomolar. Very high. The same exact high average library fragment size. Sequenced that on HiSeq and the frag insert size histogram on the right-side shows exactly the same size distribution of the insert sizes very, very consistent.

Please go to the next slide.

If you look at the genome coverage, again, 50x coverage and we have pretty much identical coverage and we only have 7 base pairs of gap in the first chromosome and 129 of the second. That's basically identical results obtained at 10 nanograms of input DNA which is really impressive, because no matter what the sample is, if you can pretty much quantify it with anything, you can sequence it using this method.

Next slide please.

Here's another example of a different Burkholderia that now goes to a higher coverage level. So this is 150x coverage of both chromosomes. The size distribution is beautiful. The evenness of coverage is absolutely beautiful. There are no large gaps. There are very few gaps, so this is very suitable for either read mapping or assembly.

Next slide please.

Here's now a Bacillus anthracis Sterne strain, and again showing that with a low GC organism, we can achieve the same results. You can see here the average fragment size is 400 base pairs. We started with 100 nanograms of input and only performed 8 PCR cycles and still obtained 72 nanomolar library concentration. No primer dimers, no adaptor dimers, very clean and nice.

Next please.

And here's the coverage plot. So you can see here that the Plasmid (pXO1) is very covered well to a very high level of course. The genome is also covered to a high level and you can see that this was a rapidly replicating genome because at the origin of replication at the left and right side of the graph we have twice as much DNA as in the middle, because the DNA was being replicated from the origin of replication away from it. Again the coverage is very, very even and we only have 7 base pairs out of 4.6 million that are not covered by this sequencing method.

Next please.

Just wanted to mention, to us this is very important, these two slides here. The slide on the left, we simply applied this method that we developed to our single-cell genomics efforts that we have here, so we sequence a lot of single cells from complex microbiomes. Here we sequenced an MDA amplified product from a single bacterial cell, processed with the NEBNext Ultra. There's a bit of an adaptor dimer that's the 126, and there's a bit of a primer dimer at 83. However, that's really negligent and can be completely excluded computationally. But you can see that the size distribution of the library is beautiful. It's sequenced beautifully and gave us beautiful data from single cells. Then another important aspect of this kit is how robust it is. The very first time that I trained a student who didn't have any experience with these kits, she was able to reproduce absolutely beautiful data with exactly the same... You can see the library size distribution is exactly the same. There are no adaptor dimers. It's just worked beautifully and it sequenced beautifully. So that's a very important aspect that this method works every single time, on any given day, in anyone's hands and that's one of the reasons we love it.

Next slide please.

We have done this extensive project with comparing different kits on different bacteria and performing different computational analyses to try to map genomes and assemble them. This is just a quick graph showing how many contigs and scaffolds we obtained when combined Illumina data, that I showed you with PacBio data for the long insert libraries. And then ALLPATHS assembly, you can see that for Burkholderia it's particularly challenging to assemble those genomes because their high GC contents, so we obtained quite good results. For example, Bacillus anthrasis on the right side, we obtained very good results. There are very few contigs and very few scaffolds. In fact, it was fully sequenced. The full genome was sequenced.

Next slide please.

Well that's all I have, so I'd like to thank very much the NEB team that helped us work through this process. It was very easy and nice to work with them. Also, we have the LANL Wet Lab team and the LANL Computational Team on the right side. And for more information about any of this, you can contact me or Patrick Chain who is my colleague that does all the computational analyses. And I have a couple of pictures from the beautiful Southwest.

Ed Winnick:
Okay, thank you Dr. Vuyisich. Our final panelist today is Daniela Munafo from New England BioLabs. Please go ahead Dr. Munafo.

Dr. Munafo:
Good afternoon everyone. Today I'm switching gears in the talk. I'm going to present directional RNA sequencing from low-input amounts.

Next slide please.

So as you are probably aware, the standard RNA-seq workflow is very laborious and time-consuming. Because it involves numerous enzymatic and clean up steps that leads to significant sample loss and reduction in library yield. Therefore making this workflow not compatible with low-input and neither suitable for automation. And in most cases, what is even more important, they lose this information about strand orientation.

Next slide please.

Standard or directional RNA sequencing has been shown to substantially enhance the value of RNA-seq experiment. By providing the information of which strand of the double-strand DNA gets transcribed, helping to identify sense and anti-sense transcripts that are transcribed from overlapping genes. And helping to demarcate exact boundaries of adjacent genes transcribed from different strands. More accurately, major gene expression or transcript abundance.

Next slide please.

So, NEB is introducing a new tool for directional RNA sequencing. This is the NEBNext Ultra Directional RNA Seq Library Prep for Illumina. This kit has an incorporated all the advantages and improvements of the NEBNext Ultra DNA Kit that has been nicely highlighted by Cynthia and Momo. We incorporate those improvements and we upstream the reactions up front from RNA isolation, fragmentation and priming as well as the double-strand cDNA synthesis. Making the workflow, combining enzymatic reactions steps, this workflow is fast, simple and more stream-lined. And what is more important, the number of clean-up steps has been reduced only to two. It makes the workflow fast. It can be then, only involving 30 minutes of hand on time, and reducing the total time to five hours, compare with two days that it used to take in traditional standard workflow. Moreover, since the number of steps transferring samples has been reduced, it make it compatible with low-input amount. This workflow is compatible with as low as 100 nanograms of total RNA, and as low as 10 nanograms of purified mRNA or ribosomal-depleted RNA.

Next slide please.

So going in more detail into the workflow as I mentioned. The workflow is compatible with ribosomal-depleted RNA, as well as total RNA. So if your start material is total RNA, this workflow can be combined with NEBNext Poly(A)mRNA Magnetic Isolation Module to isolate the mRNA on magnetic beads. And in one step, the mRNA can be eluted from the beads, at the same time fragmented, and the random primers get annealed. We are performing the solution step in a buffer that contain magnesium. So by hitting the sample, these three reactions get takes place all at one in the same step. There's no need to purify the RNA fragment, so therefore there is a reduction in sample loss. At the end of this step, the RNA fragments are annealed with the random primers and ready to perform first strand DNA synthesis. For this step, we are including a new RT enzyme provided by NEB that is M-MuLV RT Enzyme Rnase H-minus. That can increase the yield of the cDNA synthesized. For second strand cDNA synthesis, there's no need to clean up step before first strand going into second strand DNA synthesis. During this step, the buffer that we're using for second strand contains uracil and so the sample the second strand DNA kit labeled with the UTP.

We are using this method with consistent marking of the second strand with UTP and posterior exhibition of this strand after that, total ligation. This method for directional RNA sequencing has been highlighted in the literature as one of the best methods. If you are curious to learn more about the method you could refer to that paper that we highlight at the bottom of this slide. This technology has been developed by Max Planck and is now incorporated for New England BioLabs.

As I said, second strand cDNA contains uracil. The reason end-repairing step that allows to create blank ends and then TA overhangs of the double-strand cDNA. Again, all the enzymatic improvements done for the NEBNext Ultra DNA Kit has been incorporated, so I'm not going to go over... Cynthia and Momo nicely highlight the improvements that in the downstream steps. After second strand DNA synthesis, end repair dA-tailing gets done, there's no need to clean up and adaptor ligation of [inaudible 00:35:00]. We are using hairpin adaptors developed by NEB that also contains uracil. Therefore there is, before going into PCR, there is a USER excision step. During this step, the loop adaptors open as well as second strand that was marked with uracil gets degraded. So only one strand DNA goes into PCR, and this strand during amplification gets differentially tagged at the three prime and the five prime ends with P5 and P7 preserving a strand orientation information and this is what it makes this workflow of directional. And also during PCR, other particles can be introduced, muliplexing the library.

Next slide please.

So if we look at the library performance, what I'm showing here in Figure 1 and Figure 2, is the typical library profile that we obtain. In Figure 1, we are looking at mRNA libraries made from 1 microgram of Universal Human References total RNA. Since we have optimized size selection, you can see the library, it has a very nice narrow profile, with a main peak at around 270 base pairs. The library is very clean, no adaptor dimers in the primer line in dimer form. The typical yield for 1 microgram total RNA library, it's around 500 nanograms. It's a half a microgram typical library yield. On the right side of the figure I'm showing that it's also possible to go as low 100 nanograms of total RNA, as showing in the Bioanalyzer traces.

In Figure 2, we are looking at a reversible depleted RNA library. This library was made with 80 nanograms of ribosomal-depleted RNA. And again, the same nice, narrow size distribution profile with a main peak around 270 base pairs. And a library yield of around 600 nanograms. On the right-side of this figure, I'm showing that it's possible to go as low as 10 nanograms of ribosomal-depleted, or even 1 nanogram. And I'm not even showing in this library, but we also have libraries with 100 picograms of ribosomal-depleted. But we see some inconsistency at such a low level. So for the 1 nanogram for instance, we see a little bit of adaptor dimer, but that can be prevented by diluting adaptors or maybe just performing an additional clean up step after the library. We have fully characterized the 10 nanograms ribosomal-depleted library, and in later slides I'm going to provide some more data.

Next slide please.

On the screen, what you're looking at is the quality and the sequencing data generated by directional Ultra RNA Seq kit. For this purpose, we made libraries with 1 microgram of total RNA, Universal Human References total RNA, as well as some of the depleted RNA. For comparison purpose, we choose to make libraries with Illumina Stranded mRNA kit. We sequenced those libraries on a single-end read 50 base pair reads on Illumina instrument. One sample per line, and we generate roughly 30 million reads per sample.

On Figure 1, we're looking at the percentage of mapped reads. When those reads map with the ref-seq. And we see a high percentage more than 90% of the reads map with the references. Either mRNA libraries or ribosomal-depleted libraries than with RNA Ultra Kit.

On Figure 2 we are looking at how well those transcripts were covered. So this graph on the x-axis, we are looking at the whole transcriptome, how well is covered from the 5' end to the 3' end. We are looking at the relative distance across the transcripts from the 5' end to the 3' end. On the y-axis, we are looking at the relative read coverage, and so we will see is that there's a nice even coverage distribution on libraries, then mRNA libraries as well as the ribosomal-depleted libraries, than with NEBNext Ultra Kit and as is suspected, the mRNA libraries had a better coverage at the 3' end since these are single-end reads.

On Figure 3, we wanted to see how much ribosomal RNA was left after using the NEBNext mRNA Magnetic Isolation Module. What we see in this figure is that the NEBNext libraries had very low percentage of ribosomal RNA left after the enrichment. It's less than 1%. And this is very nice because we get much more information about our analytic experiment mapping roughly 80% of the RNA with the coding regions.

Next slide please.

So we wonder how robust this workflow was, and so for this purpose we investigate how well Technical Replicates correlates. In this slide I'm showing correlations between Technical Replicates and correlations between different inputs. If we look at the graph that is in the upper-side of the slide, we are looking at NEBNext Ultra mRNA libraries made with 1 micrograms total RNA, two different samples, and we're looking at the FPKM, only how well this set correlates between samples and you can see there's a nice correlation with a coefficient of correlation of 0.99. And when we look at how well the contigs correlate between different input amounts, we also observe that there was a nice correlation of libraries made from 10 nanograms of ribosomal-depleted RNA comparing with the 100 nanograms of some of the depleted RNA libraries.

Next slide please. Next.

The slide is just highlighting for the Technical Replicate graph. We're highlighting and zooming in the low-expressed transcripts. You can see better that even at the low-expressed transcripts also correlate very well. What it was nice to see is that we saw that the libraries at low-input, the 10 nanograms of some of the depleted libraries, we found comparable transcript numbers with the high-input libraries, meaning that libraries retain complexity even at low input.

Next slide please.

When we look at strand-specificity, if you are doing a non-directional workflow, as you suspected on 50% of the reads will randomly map to the annotated strand that in this bar graph this annotated strand is shown in blue as a sense. 50% of the reads will map to the anti-sense strand, showing the graph in red. So that's if you're doing a non-directional work, where's there's lost information of strand orientation. For NEBNext Ultra directional kit, what we see is that 98% of the written map with the expected or the sense strand, they're making this method highly strand-specific. We observe that Actinomycin D is required and the reason for this is that an intrinsic activity that has the RT enzymes. The reverse transcriptase enzyme doing first strand cDNA synthesis can not only extend over RNA templates, but they can also extend over newly-synthesized first-strand cDNA and if this happens, a second strand is made during first-strand cDNA and does not get labeled with the UTP. Well actinomycin D inhibits this DNA dependent DNA polymerase activity, therefore increases the strand specificity.

Next slide please.

So to illustrate the directionality and the strandedness of the NEBNext kit, I'm sharing with you one slide that our collaborators from Children's Hospital has generate. This data is an E. coli. What we're looking here is E. coli transcriptome then using the NEBNext Ultra directional kit. And what you can see is how nicely the reads map on the notated strand and there is very low background on the opposite strand. What I'm highlighting here on the squares is the transcription start sites. What you can see is overlapping mRNA transcripts at the transcription start sites. And to prove that this is not an artifact of the method, next slide please.

What our collaborators did is they expressed the P19 protein in E. coli. P19 protein can bind to double-strand RNA, short RNA, 21 nucleotides and only double-strand RNA. What we see in the P19 libraries, that these double-strands were put down and then directional libraries were made by using a direct ligation of adaptors to the RNA. This is a different method to do a strand-specific library but for short RNA. You can see how nicely the P19 libraries overlay in the areas where there are mRNA transcripts, overlay on the transcription start sites. So this is confirming by a different strand-specific method that the areas that we see overlap are actually real and not artifacts of the NEBNext Ultra method.

Next slide please.

So finally in summary I present today a new tool that NEB is providing for directional RNA sequencing with a streamlined protocol, just 5 hours total time that can accommodate low input amounts and provide higher library yields. As I show by the replicate assays, the method is very robust, highly strand-specific and which we think is very nice is it can retain diversity of the library even at low input.

Last slide please.

Finally I want to acknowledge all the people that has been involved in the development of this kit. And thank you for listening.

Ed Winnick:
Thank you Dr. Munafo. As a reminder to attendees, if you have a question please type it into the question field in the GoToWebinar control panel. We have about 10 minutes to ask the panelists questions and get their answers.

I'm going to start with Dr. Hendrickson. Our first question was, "Does full-spectrum include RNA genome applications?"

Dr. Hendrickson:
Yes. Now we do-

Ed Winnick:
Okay. Thank you.

Ed Winnick:
I'm sorry, go ahead.

Dr. Hendrickson:
No problem. We do both non-directional and directional RNA as well as small RNA.

Ed Winnick:
Okay. Thank you.

Ed Winnick:
One of the questioners wanted to know, "How do you count for the increased post-capture yield with the NEBNext, given you are inputting about the same amount of template for PCR?"

Dr. Hendrickson:
I would assume that it would have to be reflective of the number of fragments that actually get both adaptor ligated on them. So we're quantifying using Qubit or bio analysis that reflects the amount of DNA but not the amount that's PCR-amplifiable.

Ed Winnick:
Okay. Actually a question for both you and Momo. Attendees would like to know what as been your cost per sample?

Dr. Hendrickson:
Well, in my case we were given a kit to try, so there wasn't one.

Dr. Vuyisich:
This is Momo. You can actually now purchase this kit from NEB, and I believe the cost per sample is $24 for the entire kit. That does not include the Covaris tubes. Those are $2.50 a piece and those would have to be used for many other kits as well. The cost is very, very low, and then we use Biomech robots to perform the automated library prep, so there's very little labor involved in the process as well.

Ed Winnick:
Okay. Thank you. Cynthia, also, have you tried the NEB method on buccal or FFPE?

Dr. Hendrickson:
No, not yet.

Ed Winnick:
Momo, sorry. One of our attendees wanted to know if you've tried another shearing method, and if so what-

Dr. Vuyisich:
Not with this kit. We have tried to use enzymatic methods, and they just were not as consistent as this one. This is very, very reproducible.

Ed Winnick:
Okay thank you. How many putative PCR duplications in 100 versus 10 nanograms?

Ed Winnick:
To Momo. Or Cynthia.

Dr. Vuyisich:
I don't know. We didn't do that analysis.

Ed Winnick:
Cynthia?

Dr. Hendrickson:
In our case, we actually did have an increase in duplicates because we increased the PCR cycles but we realized that was a mistake and we will not increase PCR cycles in the future and just keep it at the 6 cycles.

Ed Winnick:
This is for any of the panelists. What is the cost of the NEB kits compared to the Illumina?

Dr. Munafo:
Hello? I would refer the people to just browse our website.

Ed Winnick:
Okay. Thank you Daniela. How about the cost difference between the NEBNext system and other traditional prep methods has anyone looked into that?

Dr. Vuyisich:
This is Momo. I believe that the TruSeq kits cost $50 per library, I believe, but I'm not 100% sure. I know that they're significantly... NEBNext is significantly less expensive and that's what our lab manager told me. I don't know what the actual cost is but I'm sure that can be easily found out online.

Ed Winnick:
Okay. Thank you Momo. We have had a few attendees asking about if you perform, this is for Momo... if you perform sequencing on the Ion Torrent or Proton using this library prep method?

Dr. Vuyisich:
We did. We used the NEBNext Ultra method for Ion Torrent, not for the Proton but we used the version of the kit that included enzymatic DNA fragmentation. That's the part we did not like too much. So we switched to Covaris shearing and now we're getting great results. We're getting very, very consistent results on the Ion Torrent. So that's our standard method for Ion Torrent as well, and we just got the Proton. It'll be for the Proton as well.

Ed Winnick:
Okay. Thank you. Daniela, can the NEBNext kit be used for other next gen sequencing platforms like the Solid?

Dr. Munafo:
One requirement for make the workflow directional is that the adaptors cannot be blind end. It has to have hairpin shape or either the Y-shape. Directional workflows are not compatible with the blind end double-strand adaptors like are used in Solid. If someone just can go around with the shape of the adaptors, it will be compatible. That's the only requirement for compatibility. For the directional. However, I didn't went over the talks with NEB. Also for NEBNext Ultra kits for RNA, that is not directional and that will be more compatible with blind end double-strand adaptors.

Ed Winnick:
Okay. Sticking with you, Dr. Munafo, somebody wanted to know how will these adaptors perform with products randomly amplified, say the size between 150 to 400 base pairs?

Dr. Munafo:
Can you repeat the question please?

Ed Winnick:
Sure. The question was how will these adaptors perform with products randomly amplified, say at a size between 150 to 400 base pairs?

Dr. Munafo:
They perform very well. For example, for DNA we have a size selection chart. So the DNA fragments can be size selected within a different range. Actually the ranges that you just mentioned. Ligation performs fine with these adaptors in different insert sizes.

Ed Winnick:
Okay, great. This is for any of the panelists. Have you used this method with mammalian RNA isolated from FFP tissues?

Dr. Munafo:
Yes, I have made libraries with FFP samples from humans and library was performed fine. However, we are working on making it more upstream and it does not... something not related to the workflow itself. FFP samples can be very variable in terms of quality. We're putting a big effort to understand what our limitations, that may sometimes fail depending on how the FFP RNA has been isolated. But we have successfully made FFP samples using the NEBNext Ultra kit.

Ed Winnick:
Okay thank you. Also one of the attendees noted that magnetic beads are used to enrich eukaryotic RNA but what do you suggest to do with bacterial prokaryotic total RNA?

Dr. Munafo:
At the moment we do not offer any solutions for ribosomal applications from prokaryotic. There are products available out there in the market and what we have seen is variability in terms of performance of this product. Some of the data that I share with you was done with one of the best methods and it performs very well and we have also tried the kit with other methods that do not work as well in terms of depleting ribosomal RNA and we see a little bit less percentage of mapped reads. But at the moment we do not have anything for ribosomal depletion.

Dr. Vuyisich:
I'd like to add something. This is Momo. We do a lot of transcriptome analyses of bacterial and eukaryotes and mixed samples. We have had very good results with Ribo-Zero reagents from EpiCentre. That's all we use nowadays and we expect good results and we get them every time.

Dr. Vuyisich:
Ribo-Zero is specific for removing ribosomal RNA of course.

Ed Winnick:
Okay, thank you Momo. We have come to the end of the webinar. We would like to thank our panelists, Cynthia Hendrickson, Momo Vuyisich, Daniela Munafo and our sponsor New England BioLabs. If we didn't have time to get to your questions, we will try to have the panelists answer them directly afterward. Also, please look out for two post-webinar polls from the sponsor before you log out.

If you missed any part of this webinar or wish to listen to it again, a link to an archived version will be emailed to attendees.

Thank you for joining us for this genome webinar, and we hope you participate again.

Loading Spinner