Add Multiplex Required with airSlate SignNow

Add multiplex required

hello everyone and welcome to this integrated DNA technologies webinar on sources of sample cross and NGS experiments my name is dr. Han spec and I'll be serving as moderator for today's presentation the presentation today will be given by dr. Kristina Georgia Kristina is a SAS scientist here in the NGS scientific applications group at IDT she completed her PhD in Seiler molecular biology at the University of Massachusetts Amherst where she studied viral entry and release mechanisms of sv40 virus and since then she and since completing her PhD dr. jorde has specialized in the development of NGS applications working at ion torrent 10x genomics and currently at ITP her presentation today's lasts about 30 minutes and following that presentation she will answer as many questions from attendees as possible as attendees you have been muted but we encourage you to ask your questions at any time and during or after the presentation which you can do by typing them into the go-to of the questions box in the GoToWebinar control panel um there's a little up arrow symbol on the right-hand side of the questions bar if you click that that box will pop out it gets large you can type your question makes it easier also in case you need to leave early today or you want to revisit the webinar we are recording this presentation and we will make a link to that recording available within a couple of days after the presentation you'll be able to find that on our website but also on our Vimeo and YouTube channels there's lots of great content there there's more NGS information that we also have stuff on crisper qpcr and a variety of other topics so I encourage you to check those out we get asked for the slide decks for these webinars and we do share those as well and that is on our SlideShare site which is shown on the screen um you don't need to remember any of these links we will be sharing those with you also in a follow-up email which we will send shortly after this webinar okay so with all of that take care of all that housekeeping stuff I will hand this over to Christina's we will get started thank you for the introduction hans and thank you all for calling Antipodes webinar as han said we're going to be talking about sources of sample crosstalk NGS experiments and then provide some recommendations so that you can reduce a sample crosstalk so it's a general outline what we're going to do is talk about just the NGS workflow and define what sample crosstalk is and establish a vocabulary for the rest of the presentation and then in detail we're going to discuss sample crosstalk and mitigation strategies and then finally we'll close with some adapter recommendations since that is one of the the best solutions you can use for sample crosstalk so the NGS workflow here starting with library prep you take your sample with your various library kits and you add on adapters to your libraries that incorporate sample index that will be used for identification downstream the NGS library you might take into target enrichment depending on the application you're interested in and when you do that target enrichment you can multiplex your samples in many cases so that you can essentially save money by using less reagents and even if you're not doing target enrichment many customers and researchers are using multiplexing of their samples for sequencing on their platform as the throughput of sequencers has increased dramatically in recent years it's more economical to multiplex many samples going into sequencing then in your downstream analysis you can be multiplex your samples based on the sample index that was incorporated during library prep to assign the reads back to the sample that you started with now we'll go through more details of library construction here so starting with that original sample if you'll do fragmentation so that the library insert is compatible with your sequencing platform and then you can do and repair and a tailing to to back and then add on A's to your DNA then phosphorylate the phi prime ends and so now it's compatible with your adapter so they're going to be used for sequencing so in the adapter ligation what I've shown here are what I refer to as Y adapters and these contain the P 5 and P 7 sequences that are used for clustering classification on the Illumina flow cells and then the sample indices are incorporated here as well now for Y adapters 13 base pairs are complementary between the top and the bottom strands and that includes that P overhang this used to match up today and part of the reason that you would use Y based adapters so that you get library conversion of both the top and the bottom strands so this leads to two better conversion of your sample into final NGS library so following ligation we'll do a clean up to remove excess adapter and then you can do another library amplification if you say started with low sample input using P 5 and P 7 primers so that you have enough material to go on to subsequent steps and then you'll do another cleanup again to remove any excess primer now depending on your approach you might use one of two indexing strategies so what I described on the last slide was full y adapters which can contain the sample indices alternatively you can use what most people refer to as stubby wise so these have the same indexes that are complementary the or set the same region that is complimentary the 13 base pairs but they're truncated at the read one and read two a sequencing primer region and what you would do is in your later library amplification step instead of using P 5 and P 7 only primers you would use primers that are contain your sample index so we should have different approaches for you that can be used for library construction there's benefits to each one depending on what you're interested in doing but I just wanted to describe both here and moving on to target enrichment so many IBD customers use our our products for capture so we offer lockdown probes which are unique in that they're individually synthesized and then QC before being normalized and then finally pooled so you have an equal molar concentration of all of your probes and this really leads to superior performance compared to other other products and what I've shown here is a diagram of what's happening in target enrichment so you have your single-stranded library of molecule with a universal blocker that binds to the adapter region to prevent daisy chaining more effectively oligomerization of different target molecules causing nonspecific pull down and then here we have the x-gen lockdown Pro which is used for enrichment of the target region of interest and then we include cot one this prevents pull down of nonspecific repetitive regions of the genome essentially again more daisy chaining like molecules and then to complete the capture you take the hybridized product and incubate it with drive avidin beads so that you can then pull out just your target regions of interest so target enrichment is a really simple approach that works for many applications so regardless of whether you're doing target enrichment or some other library construction method you then need to go into sequencing so in this diagram we have the P five and the P seven regions as well as the i5 and i7 indexes and then the insert from your your sample so clustering amplification is the first step in sequencing what I'm showing here is where a molecule has hybridized to a p5 primer that's fixed on the photo so that a polymerase comes in and extends and completes that library molecule that product can then fold back on itself and the p7 which is attached to the flow cell can be used for completing the next round so this is completed many times until you have many copies refer to as a cluster then during sequencing the first primer that Emile's is the read one primer and this is used for sequencing the insert as I've shown here and down here relative to the flow cell then the next primer that's used is the i7 index primer and this again is sequence towards the flow cell and then depending on your sequencing platform you can sequence the i-5 index in one of two orientations it's a lot of the first sequencing platforms but Illumina introduced so the high C 2000 2500 my seek a p5 oligo on the pro cell is used for sequencing the sample index in the newer platform like the next week and high seek 3000 4000 instead it's sequenced an opposite orientation towards the flow cell so really a big difference it just means that you set up your sample sheet differently depending on the sequencing platform that you're using so then the last round of sequencing you are using the read 2 primer to now sequence the other side of your insert as I've shown here so what is sample crosstalk what you should have in a normal case is a sample which has the addition of sample indexes which can be used during de-multiplexing to assign your reads back to one sample or another so this is the way it should look what happens in crosstalk is that therefore the mismatch between the sample rate and the example index reads and there's various mechanisms that can cause this which we'll talk about for the basically the rest of the presentation but when you have this mismatch once you do the multiplexing based on just the sample barcode you end up assigning some reads to the wrong sample but this generally happens at very low frequencies so what applications are affected so I've provided some examples of applications that are impacted by sample crosstalk so for example low frequency somatic variant detection or ancient DNA research viral detection gene expression and then finally microbial profiling and this is just a selection of examples but what they share in common is that they're very sensitive applications where you know a very limited number of reads might change the interpretation of your result as another example I'd like to point to some work that we did with Bob Fulton from Wash U a few years ago what he was interested in doing was targeted sequencing but very high sample multiplexing 96 samples per capture and what he found was that when he multiplex to such a high level and went through capture and then sequencing that he had crosstalk such that a variant that was in one sample would sometimes be found in the other sample and what we figured out working with him was that it came down to the indexing strategy that was being used and so we've developed custom dual index adapters for him so that he could do this really high multiplexing how my point is that IDT has actually been making custom dual indices for for quite a long time with our customers to help improve their assay performance but it so so crosstalk is not really a new thing but it's also a little challenging it wasn't easy to do say the ordering and figure out exactly what the best adapter was and so part of the point of this presentation is to show you that we're trying to help develop tools so this can be more standard and commonly utilized so the source is the sample crosstalk or quite numerous so you can just have contamination of one sample into another and and that kind of issue there's really not anything we can do about that you can also have contamination of one sample index into another and we'll discuss that in great detail you can have index hopping during multiplex capture as well as index hopping during cluster amplification and then on the bioinformatics side you can have misread bases within an index sequence cause miss assignment and then finally you can have a sample carry over from a previous sequencing run so we'll talk through most of these examples and then provide recommendations for adapters to help reduce those issues and to frame this we're going to kind of walk through the NGS workflow and describe how different adapter schemes can be used to adjust crosstalk so the first one we'll focus on is just the index contamination base so it's important here to step back and think about how samples are being indexed for most commercial library kits right now so most our design you know say 496 samples and what they use are eight unique I five indexes and 12 unique I seven indexes and so it's the combination of the i5 and i7 they're used for identification of the sample and so working through this in detail if you were to say go down the column here these are going to have the same I said in the sequence but they're going to have different size and if you go across the row you have the same I five but different i7 this is a really convenient strategy if you want to index a large number of samples with a low number of indexes so in this case only 20 indexes are required to multiplex 96 samples the downside is you've used every single combination of your adapters and that can lead to crosstalk as we'll talk about so we'll go through an example here so here what a diagram is for sample one there is an i-5 with a Shannon gray and the blue is used for the i7 which is once this corresponds to the well position a1 and the next sample is in a2 as shown here and to make the math easy I'm showing an example with just one percent contamination what this results in is when you do your two multiplexing when you see the a1 combination that will be represented in 99 percent of the reads when you have one percent contamination and those would be correct but in the other case where you have a two those will happen at one percent and these reads even though they originated from sample a1 would be assigned to sample a - so that's the grass that we're talking about now if we move to using a unique indexing strategy in this case unique dual matched indexes shown with for example 1 we have blue signifying that I 5 and I seminar exactly the same for that sample whereas sample 2 is green if you have the same level of contamination of the a2 adapter into the a1 well you're going to sample all combinations of the i5 and i7 indexes that are present so this represents 98% that would be correctly identified and assigned to sample a1 or for the sample in well a1 I should say and then you'll have some combinations of 1 and 2 and how these index combinations aren't expected to be present and so those would be filtered out in say your unassigned file when you do do multiplexing so these would be removed which is what you want now at a low level you'll see representation of the to index on the i-5 and the i7 and that's going to be present at point one percent so there is some level of miss assignment but it's been reduced to orders of magnitude relative to a combinatorial approach so it's a good mitigation strategy but it does not eliminate crosstalk from contamination and I'd just like to point out that index contamination can occur on any sequencing platform so if you're doing a very sensitive application with multiplexing you're susceptible to this crosstalk issue so how does this perform in with real samples so here what we did was prepare sixteen libraries with unique dual matched sample indexes and then captured them with an IDT panels in the IDT AML kantor panel and then sequence those libraries on the next week so here's what the data looks like here we've arrayed the i-5 sample index on the X and the i7 simple index on the y-axis and then just put up the representative representation of reads as this heat map and what you would expect is the diagonal so one and one shown here corresponds to this first library molecule here that's what you should see what the off-diagonal signal is are the contamination event so for example one and seven shown here is analogous to this molecule so when you do do multiplexing with the matched indexes that you're expecting 116 those would be the correct indexes but all this off diagonal field signal would be filtered out and you wouldn't see it so this is an effective strategy to remove that crosstalk from Commandment contamination and so if we sum all of the reads that are in the caused by contamination relative to what we see on the diagonal we can quantify what the miss assignment rate would have been with combinatorial indexing and it's point oh nine percent so those are services the the index contamination that can cause cross stop crosstalk what about multiplexing in your capture it shows a couple different for several different terms that are used to describe this that it could be described as index hopping swapping crosstalk spreading a signal jumping PCR so the Buddhist view out there so what's the mechanism that's happening in this case since you already made your library and attached to your adapters where's the crosstalk coming from and essentially it's coming from that post capture PCR and I'll provide a cartoon describing the mechanism for the p7 slide but the same thing can happen on the p5 it's not like one side is more susceptible than the other we're going to do you use combinatorial indexing with the same diagram we showed before after your target enrichment we do PCR with the p5 and the p7 primers now here the p7 primaries shown where it's annealed and then been extended but for some reason you have an abortive product you can complete extension now this can basically come off during the thermal cycling and then anneal to another target that's similar and then be extended this produces a mismatch between the insert and the sample index but you would be blind to it with combinatorial indexing so you would still assign this to in this case sample one even though you had a mixture of insert from two different starting molecules now in the case of unique dual matched adapters shown here if you have the same process happen where p7 is extended but doesn't complete and then anneals to a new molecule within that PCR that gets extended but now you have a mismatch between the i-5 and the i7 index so these get filtered out just like that off diagonal signal that we had in the previous example so what does this look like in real capture so here the same libraries were now captured in for 816 Plex using 500 nanograms per library and then sequenced on different runs on an XP and here's those results so there's the 1 Plex 4 8 and 16 Plex and as you can see the samples that were grouped together within a multiplex capture there is signal there is index hopping happening but again all these diagonal off diagonal reads would be filtered out when you do multiplex using the unique indexes and the the signal that you see here is actually less than what we observed with just a low level of contamination so there is info copying happening in the multiplex target enrichment but those index hopping reads are filtered out when you use your dual matched adapter and this is also quantified here for each of those so there's the point O 9 percent that we saw with the 1 Plex and it increases to right around 0.5 percent for 16 samples so it's definitely an effect based on the number of samples that were multiplex together but those reads would be effectively filtered out if you use unique dual matched indexes but you would be blind to them if you would use combinatorial indexing and that's where that spreading of signal happens where a very small fraction of reads would be assigned to the wrong sample that's another dress is what's going on during multiplex target enrichment but what a little but what about subsequent steps in sequencing Illumina has recently done a webinar and then provided a white paper as well as other online recommendation for index hopping and sequencing and part of the reason this came about is because they found that on that pattern PO cells that use their new exclusion amplification xampp' chemistry that they had higher index miss assignments and they did with the original bridging city used so here they're showing the results for a non pattern flow cell so my feet I see 2500 etc and moving to the pattern plus so you can see that there is a higher rate of index hopping and it's influenced by the library construction method so it was worse for PCR free libraries so they've explained that they think that the index hopping is happening during multiplex sequencing so in the cluster amplification process and I'll show a diagram for just one index hopping but the same thing can happen on the other side so again starting with combinatorial indexing strategies what they think is happening is excess free adapter makes it through library prep so in this case a p7 free doctor can anneal to a library molecule that's already been extended off the flow cell and then be extended now this new sort of molecular recombination product can come off of that foe cell and then feed a new well making a new product that when you do do multiplexing will have a mismatch between the in-sample insert reads and the sample index which you would be blind to with combinatorial indexing when you use unique dual matched indexes and the same process takes place now you can see that you have a mismatch between both of those indexes and those would get filtered out so you can significantly reduce the class talk in cluster amplification by using unique matched in disease they've provided other recommendations for customers what they promote first and foremost is using unique index combinations when you're multiplexing on the slow cell so with the convert the currently released kits that means you're limited to sequencing just eight samples at a time because there are only eight unique I five index is available if you another way that you can avoid cross talk is to just sequence a single sample per flow cell or purling so you're not multiplexing so I can avoid the index hopping on the flow cell also really stringent library prep protocols where you remove excess free adapter can reduce the levels of index hopping again based on the mechanisms that they think is happening on the flow cell and then fine you can prepare your libraries and make sure that you store them on minus-20 for some reason at higher storage temperatures more index hopping is observed and then finally they recommend cooling similar RNA speak libraries so that you don't get say a high expression from one sample being assigned as a low expression another I think in practice that might be a bit challenging to utilize so that addresses that that crosstalk that we observe in targeted enrichment in sequencing what about it an analysis you can have index miss assignment also referred to as the multiplexing noise it's important to consider how unique each sample index is from another so there are errors which can be introduced during synthesis library preparation amplification and sequencing you really should use a unique sequence tag for each sample and take into account those errors that can be introduced so that each tag camp is unique and can't be miss assigned to another tag so one of the tools that you can use is added metric or Levenstein distance which is a youth which is used to account for those substitutions and index which will describe on the next slide and then finally you need to avoid homopolymers match GC content exclude self compliments be color balance and consider your sequencing platform so a four color versus two color and those are the same considerations you would have for any sequencing platform you need to make sure that you're going to be compatible with your your system so here's a walkthrough run edit distance I think many people are familiar with Hamming distance Hamming distance is a way of quantifying the difference between your sequence tags when you have substitution so how many substitutions would it take for you to get from tag 1 to type 2 and if you just have substitutions it actually takes seven with Hamming distance if you use added distance this takes into account substitutions and indents so shown here is first a deletion of one base and then an insertion of another so you actually only had two changes that occurred to go from tag 1 to Tech 2 which seems pretty unlikely but consider the other case where you maybe you made just one deletion and now as you're doing the sequencing you get the next base which is aji assigned as though it's the barcode and so now you've really only made one mistake but you miss assigned that sequence tag and so what this group did is develop some tools so that you can check the indexes you already have based on edit distance so again taking into account ndele and you can also use their tools for design of sample indexes and those are freely available on github so here's a summary of some of the designs that they did in that paper so they increased the tag lengths so this is the sample index links and then also increased distance and so how many barcode combinations are available that would be unique meeting that criteria so for example if you had an eight base sample index is being used there's more than 500 different indices available that would meet that edit distance requirement but if you want even more distinct indexes say and at a distance of four now there's only a hundred sequences that are available so it's a really nice set of tools and something I would encourage people to look at even to evaluate the indexes they're already using another methodology that you can use is quality filtering of the sequencing of the index read to minimize crosstalk so in this publication what they did is utilize again unique indexes for the i7 and the I five index and then they did the same sort of analysis we did where they looked for every single index combination that could be made and then the unexpected were on the diagonal and the expected samples are sorry off stagnant when unexpected in the diagonal would be expected and so what you can do is sum the number of greens for the unexpected and then plot that relative to miss assignments and if you increase your quality thresholds on the index barcodes themselves you can actually reduce miss assignments which is a kind of a it's a really cool knob to be able to turn to improve your assay and improve or reduce your sample crosstalk so here one of the things that they recommend is kind of fine tuning this on an ad-hoc basis so maybe you increase your quality threshold too much the number of reads they're actually left with for analysis might be too few and if you're too lean in you might have too many barcodes coming through that are messy and then those cause too many missed assignments so you can this on a run specific and application-specific threshold to minimize crosstalk so very nice has tools and methodology that can can be employed so in summary four unique dual matched indexes and how they reduce the sample crust out of it I have a figure here showing first the adapter contamination that we discussed so if you have that 1% contamination you'll have the mismatched barcodes or unexpected combinations get filtered out which is good you will have a low level of missed assignment that's still left over but it's been reduced two orders of magnitude relative to a combinatorial indexing strategy you can also have index hopping happen during multiplexer enrichment again by having an incomplete extension product anneal to another molecule in the multiplex post-capture PCR and they get extended but these would get filtered out as long as there's a mismatch between the i5 i7 index then during sequencing and cluster and occation if you have excess free adapter this present that can produce a molecular recombination event between the sample and the sample index but you can filter those out if you use unique dual matched barcodes and then finally if you take into account things like edit distance you can ensure that if you do have a miss assignment of a barcode based on the sequencing error but it's pretty unlikely you're going to make that mistake on both sides and so if you have a mismatch between your indexes again you can filter those up so many of you are probably aware that alumina and IDT are partnering on NGS multiplexing and exome enrichment so what we're co-developing our proprietary index kits that will be compatible with Illumina library products and sequencers they're highly optimized for use on platforms with luminous two-channel chemistry and pattern flow cells which they're going to be using on the nova peak and we'll be able to extend the number of unique dual index indexes from the eight that they currently have to 24 and you can pre-order those kits now eventually we will expand to 96 Udi's at a later date and then the optimized index codes that Illumina has developed are available to order through IEP for incorporation into custom adapter orders another product that we have available now are dual index um I adapters which we've developed at IDT so these we generally refer to as a three-in-one different design so these have an i-5 index and an i7 index but they also contain a um eye which can be used for a sequencing error correction save for low frequency variant detection and if you're interested in information on how to use these adapters or how you can apply them to your research and encourage you to look at a webinar that we did a few months ago on these adapters which was focused on low frequency variant detection and I have a link to that at the end of the presentation so how can you use these adapters first as maybe you're doing just a gene typing application which isn't very sensitive you might use the i7 or the i-5 index for sample demultiplexing and then read one and week two for the sample insert in the second run mode for more sensitive applications we'd recommend that you sequence both eye five and the i7 index in addition to your insert so that you can reduce sample crosstalk and then finally if you're doing things like rare variant detection you can sequence the umi downstream of the i7 index by just adding a few cycles during the sequencing so this is where the three-in-one solution you get one adapter and you can three different ways depending on your application and and what you're trying to do so conclusion all sequencing platforms are susceptible to crosstalk for multiplexed sensitive applications a unique dual indices mitigate sample crosstalk and enable sensitive applications an IDT has 384 pre-designed unique interface for custom adapter so they have an edit distance that's greater than or equal to 3 they're color balanced two sets of four and design for both two color and four color sequencers with 50% GCM if you have any questions about these adapters or maybe help with a custom design maybe you use a different library construction methodology I'd encourage you to reach out to our application support team and they can help walk you through design and ordering and all of that so there's a lot of information out there on sample crosstalk and index hopping so I wanted to provide you with summary of the publications that I referenced it's pretty much impossible to go through all of these in detail and just a 30-minute webinar so these are here for future reference so with that I will thank you guys for your attention in it for questions okay so before we move on to the Q&A portion here I just really want to capture something really quick in a poll question we're very happy with our partnership with Illumina and I just want to know if people are interested in getting more information about that and upcoming products so if you could just answer the poll question really quick um as far as the additional resources we will make the slide deck available to you all it's already posted online and you can see the link to that will be in the chat box window and we'll also be sending you a file email with links to the slides and whatnot so you can get that slide deck look at those additional resources and references as well as pull up the contact information on that conclusion slide for our application support so I encourage you to do that if you have other questions if you haven't already and entered your questions in for this webinar just type them into the questions box and they go to webinar control panel and Christian Christina and I will get to those shortly just give everybody another moment on the whole question [Music] okay I think that's enough time I want to make sure that we get to as many of these questions as possible so I'm going to close the poll thank you very much for your responses everyone okay we do have a few questions already Christine so let's get going on those all first one was an early question it was just basically asking you to explain all the crosstalk occurs but the rest of the presentation is that so I'm going to move on from there um the next question is do we have data for the combinatorial vs. compared index adapter so if I go back I think this will address the question I hope whoever asked if it doesn't please enter it in the chat again so to do save this analysis we used unique indexes and then when we did diplomacy flexing we think will deflect for every single combination that's possible even though we were only intentionally testing matches of one to one and two to two and three to three Center and so all this diagonal diagonal signal that you see would have been miss assigned if we have used combinatorial indexing but we wouldn't have been able to measure it because we would have been blind to it so essentially if we had used just the i7 index then all of this signal would collapse down and so these would be same incorrectly assigned to example two and then the same is true for if we had just used the it--i've index does that make sense so it would be just collapsing it this way I believe so if they it looked like she said if you do have a need for further clarification I just type it in and try and come back around to it yeah on the next question that we have is what is the level of contamination that's possum from previous runs it's an interesting question that is an interesting question so this is something I've heard about more from customers so I don't have first-hand experience with this and I've expanded publication but I think it depends on the sequencing platform and I think that you would need to run different indexes on different sequencing runs and then in your sample sheet look for so say you're used index 1 1 in run 1 then in run to do not use that sample and run the other indexes and then in your sample sheet still do multiplex as though sample 1 1 was there and see how many reads are carried through and you could do that on several runs does that make sense but we haven't done that exact experiment but a charges will be instrument specific and it's hard to make a general claim sure are the publication's that you shared are those all on the Illumina platforms that are there are there publications on this issue on other sequencing platforms oh so the Edit distance paper so edit tag they evaluate indexes from 454 and Ion Torrent I believe and I think they did those evaluations in silicon but they might have also made I know they made like 3 different libraries they're for different purposes I just can't remember if they were all Illumina or not and then I believe the microbial paper you used Illumina and other sequencing platforms but I'm not a hundred percent on that ok and it's also hard because to my knowledge I am Tauron uses a single indexing strategy ok this next question I guess I don't um I don't understand it completely maybe you'll understand it so I'm going to ask it if you don't know it we can ask you some further clarification so how many cells can be used for the custom dual indices and will it be okay for more than 500 cells I think they're asking maybe about RNA seek applications but I'm not sure okay let's just move on but if you want to clarify that question just type it into the question box and if you can answer your questions I think if the question is can you have more than 500 indexes like sample indexes available for the like the three-in-one design I talked about that one is based on a neat base their sample index and if you want more than 384 you need to make the symbol index longer okay you have more combination possible here's another good question so with the um I and the sample barcode in the adapter does the adapter become does the additional length make it more likely that the adapter will form adapter dimers Oh at least in our experience Melva testing we've done no and the umi is very diverse and so maybe they're worried about like enrichment or something where you get over representation of one molecular of our code or something like that and we don't see that okay although this next one is a configuration question so are the 384 unique indices offered used to make a total of 384 libraries or are there actually 384 left and right indices so I am going to be a little bit confusing here so it's there are three hundred and eighty four pairs of the and you can kind of as long as you don't reuse an i5 or i7 index on any other samples they're unique so there's 384 unique combinations okay for 1884 samples do we have a clear date yet for delivery of the UT is um I think the first set are supposed to be later this summer the first set of 24 but I don't have a firm date on that now but it's coming fast we'll have those soon yeah it's actively being worked on right now are we doing anything custom right now until that's like a final product yeah absolutely and we have many customers who need a solution now and so we're working with them to make custom adapters and so the sales support team can help with that or the sales applications application support team sorry can help with that and help you with recommendations for like what modifications you might need or or purification that can be because it is a little complicated and so we've tried to part of the reason part of the rationale behind the tool index EMI adapters is to streamline ordering because it is a little bit complicated and so we want it to be as easy as possible for our customers to get what they need as soon as possible sure okay so someone has a question about slide 15 and they wanted to know why is the crosstalk signal in the diagonal in the lower right half why is it not why not adjacent or random oh oh oh they want to know basically I think why why is it crosstalk was the same basic pattern yeah so actually they are adjacent physically so the one one combination was near the 7/7 combination when the pair was arrayed in the plate and then duplex but you just can't see that here so you'll see that even spacing is because this one and then this one and then 13 here those three were actually in the same row in the poly model so there's a physical relationship here exactly it's so similar to what I'm showing here where you're more likely to get wrote are well a 2 in a 1 then you are from a 2 to fac-2 gender stick yeah so there is it so and that's what kind of points to it being maybe an operator just you're opening a plate in using it multiple times that kind of thing can cause that contamination and make it worse okay here's another one from somebody about they want to know are the unique dual indices more beneficial or equally beneficial than the unique matched dual indices so I think there that's what I'm wondering if there's a distinction so I don't think so like for the purposes of the cartoons they made it was much easier to have a matching color on both sides but I don't see any reason why you couldn't adjust your design to say have the indexes not match as long as they're unique and not reused on another sample then that's fine so all the stuff that we have all the data presented here did have matched adaptors but there's no reason they couldn't be miss master staggered okay this next person's asking you for it educated yes here you ready they want to know what our what are the reasons as to why you might have index hoppin and poppin happen at minus eighty degrees no idea um I wonder if they've seen it yeah I'm a little confused like that's because the data that they showed it and I would strongly recommend that people look at the the webinar is it Illumina did they had a lot of nice data in there showing storage conditions and things like that but obviously I didn't get into because we didn't do that work but yeah they have something very interesting data where libraries were multiplexed together and then store it for degree use versus minus twenty and for some reason they saw more index hopping but I don't know what the mechanism is okay um next question is can we mix the classic can we mix the UDI indices with the classic adapters so can you incorporate UTIs in with the classic adapters I guess yes with Y adapters yeah I mean that's basically what we used for say this experiment we're classic Y adapters with just unique indexes for all 16 samples and you can also use the unique indexes when you're doing a PCR based indexing so the sub e wise you can also index in that approach and that works equally well as far as we can tell we've had collaborators test it and they've seen no difference okay perhaps a good follow-up question of that is with the addition of the um I what is the level of accuracy or error percentage in a multiplex compared to the UDI ah I'm not sure if I can you see that again you don't think I understood the question so regarding the addition of the UM I what is the level accuracy or error percentage in a multiplex experiment compared to the UDI so adding in the UM I doesn't have an impact on the ability of the simple unique combinations of the sample indices reducing crosstalk so Sigma is sort of this separate piece that can be used in your bio informatics pipeline to do error correction of the sample read so that you can improve your your accuracy for low frequency variant calling but it doesn't affect crosstalk but if you're doing something with um eyes and you really want to do a sensitive application you really should be using unique sample indexes because of crosstalk like if you're trying to detect a 1% or sub 1% variant then you should be using unique indexes because it's exactly what we've been talking about if this these reads here were all collapsed down to sample two now you have significant crosstalk such that you would get maybe false positive holes you know maybe false not negatives in the case of a tumor normal analysis that kind of thing it's okay pretty bad measure your begin those indexes for a long time like I said we help us call them years ago so we kind of had a solution but didn't realize how much of a problem it was for so many people yeah and then this is actually I guess to get follow up to what you were just saying are the UM is completely random the yeah so the molecular barcode the um I those are nine based on worse they're just degenerate spaces so that yeah and what you and this is coming from not an expert perspective here but what you want to see with the UM eyes you don't want to see one repeat on two different things right yeah yeah and so we have significant diversity there are so arthas nine raise fares for the applications that we've been focusing on like low frequency variants detection we found that sequencing just the first six base pairs is sufficient to have plenty of diversity for your application sure and so like if you find a rare variant or you think you have but they both have the same um I you know that there's a issue there that there's probably an error if I understand that correctly ah so so it's kind of a bit I don't want to blend it too much and I would definitely encourage people go to zero dose later yeah when you're using um eyes for error correction what you're trying to do is tag your molecule and then sequence that molecule with that tag many times and then you can build a consensus for all the times you read that molecule that's really accurate incorrect you you actually want to see the same variant let's say it's many reads of exactly the same library molecule with the same PMI if it were only in one reason maybe it is a sequencing or something like that do that make sense yeah I think so yeah visit err correction way of using molecular barcodes there's other ways that you can use you I okay next question and we're getting pretty close to the end of time we're not going to get through all these but we'll get through most of them the next question is do the UTIs arrive as pairs when you purchase them I believe they will and if you are say ordering the adapters now is a custom product we can do we can do Plex them so you plug feminine and i-5 index together and provide them in very just you know various different products could be in tubes or in plate serve kind of whatever you need [Music] okay um so I just want to just kind of catch everybody up here really quick just one more time about resources there's a couple of questions about slides versus the link to the webinar so in the chat box there is a link to a presentation by Myrna Yash and she covered the the UM eyes for sequencing rare variants that is a great presentation there's there's a really nice animation of how that is used in the data analysis so that's great and then if you just scroll up a little bit there's also a link in there to the SlideShare site where the slides for this presentation are already uploaded so we can you know you don't need to go look at those things now but that's available to you will also send you an email with those links so yeah I just want to be sure we discuss that really quick um I see uh one more question here um this question says ah I guess I don't really understand what they're asking for Oh with pee pee eighteen and nineteen what is plex referring to I'm guessing that's slide 18 and 19 the word Plex oh so is it short for multiplex so one if I don't have an individual sampling each capture verses four Plex is four samples we'll put together and then captured at the same time okay um all right so it's 201 we still have a few questions left we're not gonna be able to get to everything uh we will follow up with you it may take a day or so but we'll get back to you about any unaddressed questions and then near the very end of this slide deck which like I said is available to you will send you links also you can find the application support email address if you want to order something custom now or if you want additional information on anything you can just email them and get information and also that slide with all the linked resources all the referred to references for this presentation are on there as well so I encourage you to go look at that if you want more information and yeah we will get back to you all right thank you very much Cristina thanks so much for this excellent presentation thanks very much thank you all for calling in