Founder Stories: Oliver Raether, Bruker & Trapped Ion Mobility ToF Mass Spectrometry

Welcome to the second webinar organized by the ETC, Education and Training Committee. My name is Giuseppe Palmisano. I'm an associate professor at the University of São Paulo, and today I have the pleasure— I'm co-chair with Justyna of the ETC, and today I have the pleasure together with Ben Collins to, to host Oliver, uh, for this, uh, second ETC webinar of, uh, 2024. So just a few slides before starting the, the webinar. So the Education and Training Committee is a committee under UPO, the Human Proteome Organization, and actually the main aim of this committee is to organize workshop course tutorials and webinars like this one that you will listen today. And we, we aim actually at joining all the UPO members and all the proteomics community in and learn about cutting-edge topics. So just one slide, I would like to advertise UPO 2024 that will be in Dresden, Germany. And please visit the website there is a very nice scientific and social program. And by the way, the first day of the UPO 2024, the ETC will be responsible for organizing pre-congress training courses that are courses dedicated also to different topics for, for young researchers, students, and so on. So by saying that, I will stop sharing my screen And I will actually now pass to Ben. Ben, the floor is yours. Thanks a lot. And thanks a lot, Oliver and Liv, for—

Thanks, Giuseppe. Thank you. So I'm, as Giuseppe said, I'm Ben Collins from Queen's University Belfast. And welcome also from me to the second webinar in the current season of the HOPO-ETC series, where we're aiming to take this deep dive on, uh, technical, technical background of some of the technologies that many of us are interested in. And so today we're very honored to host Oliver Rather from Bruker Daltonics in Bremen, where he's a research and development manager. He got his MSc in engineering from the Hamburg University of Technology, and over the past nearly 3 decades, he and his colleagues have developed orthogonal time-of-flight mass spectrometers And importantly, since 2010, this includes the TIMS TOF product line, leading to many scientific publications and patents. And relevantly for today's audience, actually, Oliver was also awarded the HOPO Science and Technology Award in 2020 for contributions on the TIMS technology. So I think we're very lucky today to have with us one of the leading experts in the field. And I can confirm actually that he's one of the, really one of the experts, because as a part of this DIA-PASIF collaboration we participated in with several other labs, we, I got to visit his lab in Bremen once and he showed me actually one of the very early prototypes of the TIMSTOF instrument standing in his lab, which was cool to see. Good, so I think we're aiming for roughly a 45-minute talk and then we have some questions and answers after this. Martin's, sorry, Oliver's topic today is the past, actual, and future directions of trapped-ion mobility time-of-flight mass spectrometry. And with that, I would pass the ball to, to Oliver. Oliver, if you can try to share your screen again. Yeah, yeah, it looks good. So I will switch off my video and enjoy the talk.

Thank you. Yeah. Many thanks, Ben, for the introduction and many thanks to the ETC for having me here. It's always a pleasure to present something at the JUPO. I had some opportunities in the past to do so at the conferences. Yeah, my title of the presentation is here. I will, I need to locate my, yeah. So this is the more detailed agenda for today. As Ben mentioned, it will take about 45 minutes, 3 chapters, let's say. First, I will go through, let's say, the TIMSTOF trapped-ion mobility separation instrumentation and give a short overview of what had been released already. And so if you are a TIMSTOF user, you can maybe sort where you are and what you can do with your system, especially from the from the different DRA techniques we have over the time implemented on the TIMSTOF and also for the future instrumentation we are currently developing. Yeah, I, for those of you who can stay until the end of the meeting, I will explain the picture you see here on the right towards the end of the presentation. So Starting with the TIMSS-MS, I have in the whole presentation not much formulas or theory about mobility cross-section or so. There are many good presentations out there and articles published, so I try to keep it as simple as possible and give you more the big picture, let's say, of what is possible and not diving into technical details. So this first set of slides is maybe the most technical one, and what you see here is a cross-section of the TIMS analyzer. So we have here the analyzer section, we have a capillary where the ions from the source are introduced, and we apply a repulsive electrical field. So here you see the electrical field strengths, and above the zero line means they are repulsive, so ions are stopped in the electrical field here. And the gas is pushing the ions towards the exit of the TIMS analyzer. And if the electrical field strengths and the gas force are in balance, then ions are trapped. That's where the name a trapped ion mobility analyzer comes from. And now there's a short animation. So when we accumulate ions, we let them introduced and you should see now that the bigger ions are stored here and the smaller ions are stored— ah, sorry, I have to switch on the laser pointer so now I can point it. So these are the bigger ions, the smaller ions, they are located along the axis of the analyser and we can now modulate the electrical field in order to release ions in basically a very flexible manner. So for the elution of the ions, we would now turn down the electrical field, and then in the mass analyser we see a peak, mobility peak, yeah. So this stands for a certain mobility, this peak, and after a time we can release the second one, or we can scan the electrical field strength down, and then we would get this mobility peaks of the species, formally stored. And as we in the TIMSOFT have of course also a time-of-flight analyser, a quadrupole for filtering, and a collision cell for fractioning, there are a lot of possibilities what you can do. I will go into this later. For now, that's enough, and I want to show you how it all began. So there had been an initial publication from Francisco Fernández-Lima and co-workers in 2010, where the first TIMS mobilegram was shown. And this period of time where we had this, this breadboard or first prototype, and until we launched the first TIMS TOF product, we upgraded the Bruker Impact Q-TOF with TIMS analysis. And that's been like 5 upgrade kits we provided to collaborators like Francesco, but many others. And in this period of time, it's like 5 years, we developed already a lot of methods for the TimsTOF. And one is maybe the most prominent one for the proteomics community is this PARSEF publication. PARSEF stands for parallel accumulation sequential fragmentation. And this is published by Florian Meyer from Matthias Mann Group in 2015. And this was really a proof of concept paper. So what you see here is not that important. Here you see the mobilegram and it was very manually programmed. So we just could address, let's say, How many are these? Like 4 different or 5 different precursors in a row, but only this because everything has been manually programmed. And this was just to envision that later if everything is automated, having a precursor selection algorithm and things like that, that it might be possible to do proteomics here. Here's one other example, also very early before the launch of the first TIMSTOF, that is from Christian Bleiholder group from Fanny Carolin Yu, and they got like 2 of these first TIMS cells, and they did run the first tandem TIMS experiments. Oh no, sorry, sorry, no, this is CIU analysis. The tandem TIMS comes 2 years later. So this is CIU analysis of ubiquitin, and Christian always did a very good job with his with a team to demonstrate that TIMS is very comparable to drift tube ion mobility spectrometry. So really in these early days people had been very skeptical about this technique, but also from the theory and if you understand the physics, it's pretty much the same than a drift tube. Yeah, and then in 2016 we launched the first TIMS TOF called TIMS TOF and that has now a dual TIMS analyzer. So to enable this parallel accumulation approach, we have an accumulation region and analyzer region to scan out the ions, as I just mentioned. And with this parallel accumulation, we just could double the duty cycle. This instrument generated then more bilograms like this, and going back to the 2010 mobilegram, so it's the same speech here, tuning mixture, and you see in this 5 years of development, the mobility resolution has been improved, but you also see that you now here see the inverse mobility. So there's a calibration applied, and yeah, so that's the progress we made, let's say, in the background without having any system on the market. And that's maybe some of the take-home messages of this webinar, that from the idea to an instrument being sellable, there's in this time a huge time. But yeah, you will see in the course of the webinar that it always takes a lot of stuff to be done before the final user can use it. Yeah, so In this case, it's like 5 years the development took place. And with this first instrument, yeah, we could do a lot. So for example, this was just a measurement of isomers, tree sugars, and you can see on this, the TIMSS is very flexible. That's one of the advantages. You can decide by the ramp speed, yeah, how fast you lower the electrical field strength. You can decide on the resolution. So here we demonstrated that really with a resolving power of 20 to 40, it's not possible to separate the 3 sugars in mobility. But if you scan slower, then you see them separated. There are other examples of isometric lipids. And yeah, what was most important is that we also provided data analysis software, let's say a viewer on the data where you could see the so-called heat maps where you see the mobility, the mobilegrams, and see the map which shows mass versus mobility. I will show that later. And we decided for an open data format. And from my perspective, this was really one of the most important decisions taken because the processing software, especially for proteomics, is really a huge effort to add this additional dimension of mobility into the existing algorithms. And many groups in the world, I think in the first year, more than 40 times our SQL or TIMSS development kit, software development kit, had been downloaded. So there was a huge interest in the bioinformatics groups, and that was, I think, part of the success that we could launch a solution for proteomics. And yeah, actually the TIMSOFT was kind of successful because it raises a lot of interest, but I remember on the launch event, most people I spoke with, and that'd be like 30 found it interesting, but they would not purchase it because the fully automated software is missing. So no algorithm to do DDA, no post-processing. So it was, I would say, moderate successful without these tools. But still, a year later, we could launch the TIMSTOF Pro that had an improved TIMS analyzer, so higher capacity, namely and we had the acquisition software to do DDA passive. So we had a precursor scaling included and we could steer the TIMS analyzer synchronously to the TOF and really decide about several precursors per TIMS scan. So we had a very high MS/MS data rate of above 100 Hz. That time that was really new. And we had also some collaborators who made it for the launch, that was at HUPO 2017 in Dublin, who made it to have a running post-processing software. So here, PEEKS and Max Quandt from Jürgen Cox had been the first software tools to analyze the TIMSSOF data. Yeah, and also then we could release the second passive paper, again Florian Mayer and coworkers. And yeah, here we could really demonstrate like the protein IDs and which was pretty amazing that time. I know that meanwhile we have higher IDs on the same sample, but that was at least for us was groundbreaking performance with this technique. And yeah, so, and here reads a note, it was about 3 years after the technique publication where we manually programmed it. So it took at least also 3 years to implement the acquisition and the processing. Yeah, and from this I can go a bit faster through it. In 2018, we launched the DRA passive acquisition method and also a first processing software that was then at that time open source. So Matthias Mann formed in collaboration with Rudi Abersold Lab and Hannes Röst from University of Toronto. And Hannes and his team really could adapt their open source software at the in mobility dimension. I remember that Anija did really a good job to implement all the new stuff into the open source. And so we could launch then at Hupo 2018 this DRA passive workflow. I have here a picture of the EvoSAP 1 because that also came out in this year. And together with the team of Oliver Worm, we adapted the EvoSAP also to the TIMSTOF. So that was another hardware launch then, let's say, for the TIMS2 family. Yeah, and here now I have the citation for the paper I mentioned already that Christian Bleiholder in the same year had the first publication with the tandem TIMS. So he put together with colleagues from Bruker Research in Billerica, he put together two TIMS in a row and could do CID in between. And there are a lot of papers out. If you look at, if you're interested, you can look at Fanny Liu. She published, yeah, I would say 5, 6, 7 papers about the tandem TIMS and what you can do with it. They have also now a second generation of tandem TIMS where they can do UVPD in between. So it's interesting, but maybe not that interested for the proteomics community, therefore I don't dive into that. Chronologically, a year later we integrated the ion source, the MALDI source from the Bruker Ultraflex and RAPIDflex into the timsTOF and offered the timsTOF then with dual source. So, we could do MALDI and electrospray actually at the same time, simultaneously. There was not that much use for doing electrospray and MALDI at the same time. But you can, without any mechanical change of parts, you can switch between MALDI and ESI, and the performance of the timsTOF LEGS is the same, it had been always the same as then for the electrospray version of the timsTOF. So, yeah, it was very successful from also for Kruker, who is the major vendor of MALDI equipment. It was a very successful MALDI product from that moment. And yeah, I have here just one of the first papers or the first paper that is from Jeff Spraggans from Vanderbilt University, Richard Caprioli group. And they really provided the imaging software and programmed the software on their own. Rüger could at the launch not show such, such images, but what you see here. So this is kind of interesting. This is a lipid, isomeric lipids having exactly the same mass but different mobilities. And you see how the tissue image changes. Yeah. So for example, the one lipid, isomeric lipid is located only here, whereas the other is more located here. And that is really a new capability to the world of multi-imaging. So that was an important milestone. And yeah, from that, then going a bit quicker through it, we have the TIMSS TOF Pro2. So Pro, by the way, stands for Proteomics, and that has the new TIMSS analyzer, and it has also a new instrument controller, which is very important in the for the workflows I will show in terms of DAA. So we could connect the quadrupole and the TIMS in one FPGA and run it really synchronously. That was important to further improve the duty cycle but also to enable additional workflows. And also in the same year, we launched the MALDI-2 possibility, that is post-ionization with the second UV laser, and which increases the sensitivity for apolar compounds like vitamin D or other special lipids. Yeah, with that, we can go on. I just want to build up this tree so that you see what the different flavors of TIMSTOF look like. So the TIMSTOF-SCP is like a third branch of TIMSTOF. It comes with a little bit different form factor, but it includes an additional differential pumping stage. So we kept at the source, we could introduce more gas into the system, pump it away in a so-called single cell cartridge. So SCP stands for single cell proteomics, and it increases the sensitivity of the timsTOF Pro II by a factor of 2, which is huge if you want to go to really lower sample amount. And then finally 2020, so we are close to today, we had a next-generation TIMS analyzer and introduced it in both systems, in the TIMSTOF LEGS and also on the TIMSTOF, which is now called HT, with that it has a new digitizer. So both systems have a new digitizer which increased dynamic range of the of the detection system, both on the digitizer side but also on the detector, because we can now amplify more in the digitizer and less on the detector, and that releases, let's say, the load from the detector. And yeah, for the MALDI, there was an increase in spatial resolution from 20 micron to 5 micron, which is also, of course, important. And also needs a high sensitivity system because the smaller the pixel is, you acquire the more ions you need out of it to get a clear picture, let's say. Yeah, and then finally last year we launched the TIMSOF Ultra. It was a further development of the TIMSOF SCP, so again higher sensitivity And yeah, with this, the TIMSTOF family as of today is complete. And what I want to point out is that Bruker has always a pretty, I think, user-friendly upgrade policy. So in principle, you could upgrade a TIMSTOF from 2016 to 2024. It's here a couple of steps and I think meanwhile we discontinue upgrade packages here in between. But if you had this old system and always purchased, of course, the upgrades, then you could have still a state-of-the-art instrument. And that is true for all branches. There are some restrictions here on the MALDI side. So the first timsTOF FLEX systems could not be upgraded to MALDI 2 because the window for the second laser was missing. So there are some technical boundary conditions where you could not upgrade, but we always have the policy, and I appreciate that our management agrees to it, that we could upgrade customer systems. Because at least for me as developer, it's leading to a better collaboration and interaction with the users. Yeah, that's the first part, just giving an overview, and now I will start again with an overview for DIA passive or generally data-independent acquisition and will explain you some new developments which are close to be released for the DIA on the TIMSTOF. So first of all, I want to point out to this publication. It's from 2020, a group of Chinese authors around the Westlake University, Hangzhou, and they really assembled all the publications around DIA and separated them in DIA acquisition publications, software tools, and analysis tools. And I have to update this a little bit because in the past 3 to 4 years there is happening a lot, let's say, in DAA. I would say DAA has become more like a standard for proteomics analysis compared to DDA. And yeah, looking at this, first of all, I added here like 5 yellow dots. Yellow means the software is supporting the Bruker TEMSofPro. Yeah, it says here in the publication TEMSofPro, and you see that meanwhile all major software tools, especially DAIEN, Skyline, and also DIA-EMPIRE, PEAKS I mentioned already, it has been from the beginning, all software tools, let's say third-party software tools which are not related to a certain vendor, then of course Bruker TIMSOFT is not supported. But yeah, this is really well accepted, let's say, in the post-processing developer community. And I add another two systems, software tools. One is MACSDA from Jürgen Cox's group, published in 2021. And then I want to add the Bruker vendor-specific tool. It's based on Dian and it's also available of course with the TIMSTOF. And on the acquisition scheme side, I also want to add one publication, especially because it's adding here on the very top, and that is I think a very remarkable paper from Shiruki Haugland-Heiser from David Klemmerer lab. And she did already 2000, so 3 years before the first, or let's say then the second DRA publication came out, she already demonstrated parallel analysis of, in this case, I think it's 5 peptides So it's a 5-peptide mixture, and you can read out here the fragments of each peptide in these lines. And this line here is the mobility, and this line is the mass. So I added it for the reason that this is exactly how the TIMSTOF is operated today, yeah. So this is what we would now call a heat map, and here's the drift time and flight time, but it's basically mass and mobility what you see here. Yeah, maybe some remarks to the other milestone presentations here. So Shotgun CAD is really one which is very important, and it's from Purvine and David Goodlett lab. The DRA from Venable is, so he's from John Jade's lab at the Scripps Research Institute. And yeah, this is a publication from Matthias Mann group. SWOS is maybe the most prominent DRA or most known DRA method developed at the ETH Zurich from Rudi Abersold group. And yeah, so SONAR is from Waters and so on. Scanning SWOS is more SciX now, and here we have the Bruker publication I just mentioned, the DIA-PASSIVE. But what I want to do is I want to sort these, these different acquisition techniques a little bit, and I marked them by colour here for you. And first of all, I mark all publications which do full mass DIA, I call it, so that which have no quadrupole isolation, they just apply collision energy So at FRUKA, we call it broadband CID. So you just switch collision energy on and off and then try to analyze the data in the post-processing. So these are these three publications which, which are doing this. And then the majority of publications are about what I call selected mass TIA, like this was where you use the quadrupole to increase the selectivity. So you step through or scan through a quadrupole window and get higher specificity, higher selectivity, and therefore can analyze more complex samples. And these two basic or fundamental differences can be expounded and could be, let's say, complemented by ion mobility. And so I have two new colors here. And if you look, this very early paper I just mentioned and the Waters HTMSE, they make use of mobility separation but without quadrupole usage. And really at that time, and I think it's still the case, the Bruker had been the first group, and together with Matthias Mann, who combined both mobility separation and selected mass. And that is, as far as I know, still really a differentiator of the DIA on the TIMS TOF. Yeah, so this quick summary, why to add the mobility separation to DIA, it increases selectivity, it reduces the FDR when you make use of the collisional cross-section information. So for each species you detect in the DIA dataset, you automatically also get a mobility or collision cross-section information. And at the beginning, also Matthias Mann group, there was a publication, they just built up a library where I think it was more than 100,000, maybe even more than a million. I think the goal for Matthias was to really have a million of peptides in a database with collision cross-section. So like a spectral library, including collision cross-section or mobility. But meanwhile, I think that is no longer used because the tools like DIANE with the neural network training, they can predict the mobility pretty good. They predict actually also the spectral libraries. That means spectral library means you have different intensities on the fragment, and that is important for library matching in DRA. That you know the probability or the intensity ratios of the peptide fragments. And that is really one of the major differences today. I think they call it direct DRA. I don't know the term, but Diane, I think, was first to introduce it, this direct DRA, and Spectronaut has it, and I think many other tools will also have it. That you now use a neural network approach and a huge training dataset. And then you can just by the sequence of the library peptides, you can in silico predict the spectra. And that means that you these days no longer need to do a DDA with maybe LC fractioning, so huge effort experiment to create your library. So that's really also a game changer, let's say, on the DRA acquisition. Yeah, and with the DRA passive, which I will explain in the next slide, we also increase the duty cycle compared to the selected mass DRA, and then I will explain how this works. So in the rest of the webinar here, I will use this heat map only. This is just more simple, let's say. You see the mobility on the y-axis, the mass on the x-axis, and typically these are the doubly charged, triply charged, and highly charged peptides, and this is singly charged species, also contain of course a lot of peptides, but normally we target here this region. For proteomics. And here in the webinar, I simplify the TIMS operation. So the red line is now a symbol, let's say, for the electrical field strengths I mentioned. And yeah, what we do is we scan the electrical field strengths and then we elute, let's say, row by row, the peptides and they undergo, let's say, collisional fragmentation, CID, collisional dissociation in the collision cell, or selection by the quadrupole. So the TIMS scan is now just depicted like this. So we have a starting point and we have an endpoint for the TIMS scan for the electric field strengths we scan, and a typical scan time is like 100 milliseconds. It's very flexible, as I mentioned, but But just here to make it a little bit more practical, I put in this timing here as an example. And if you now want to do DRA on the TIMSTOF, I overlay here a region, light blue is now, this is by the way from the control software of the TIMSTOF, the editor where you can select the region you want to address, and in this case if you want to address this region, then this is the GA passive, works like this. So you use the quadrupole to select a mass range. So a mass range or a 25 Dalton mass window is depicted here in green, and you see the red scan range of the TIMS. And if we now start the TIMS scan in slow motion, Yeah, we address the peptides on this red line. And once we reach this point here where the blue program area ends, the quadrupole jumps to here and scans the next and the next. Yeah, and then we can— so this is now 100 milliseconds, and then we can continue with the next 100 milliseconds addressing the next part of the error and so on. Yeah, so I think the animation speeds up now, yeah. So in total it takes us 800 milliseconds to go through all of these regions. That is a typical method on the TIMSTOF. And but that means it's compared to the selected mass DRA without mobility separation, it's 3 times faster because we jump, as you remember, we jump 2 times and in 100 milliseconds we address already 3 25-dB mask windows. So we have higher duty cycle and we have the additional benefit of the selectivity I explained with the predicted mobility. So this is the technique we introduced first. And meanwhile, there are a lot of groups working with this, and I want to highlight some of the publications coming out with this technique. First one is a pretty recent publication from Patricia Skrovanek from the Matthias Mann group, and she and her coworkers had provided a Python tool which defines the window size, so the window widths, based on the density. You're here on this figure from this paper, you see the density of the I guess it's HeLa species here, and the window size is broad here at the beginning where there are low, smaller number of peptides in this region, and then the window size shrinks down here to the middle, and then it becomes broader again. This gives you an additional duty cycle approach. Then I have a presentation here from Vadim Demichev lab Lukas Cheville. This is also most recent publication on bioRxiv 2022. They called it SlicePassive. The good thing is if Vadim is involved here, you always get also a DIANE version which can handle the data acquired and this is kind of optimized. And I have a couple of other examples here. This is from Nikolay Slavov lab, Jason Dirks, and he is he's doing kind of interesting technique doing multiplexing on DIA. So they use like 3-plex Amtrak reagents here. And yeah, that is also a new approach. And again, I think Vadim is here, yeah, on the list of authors. So you get directly an adapted version which can retrieve the or make use of the correlation of the differently tagged peptides, yeah? And that is of course a benefit. Yeah, another publication, SpeedyPASEF from Markus Walther group, again with Vadim here on the list of authors and also Lukas. And yeah, with this, I want to switch gears and show you something new and at Bruker we programmed already the acquisition scheme or had programmed it since 2020 and maybe you have not heard about it right now because it's not yet let's say usable for the end user so for the proteomics community. What we change is that we go from this acquisition scheme to that one, yeah. So we can now scan parallelograms and program parallelograms. We could even program trapezes, but currently we are using the parallelogram because the calibration is for the moment only provided for this kind of scanning. And if you want to address such a area here of the heat map, then we again add the quadrupole isolation window, but we don't scan now in this direction and have the jump, but we move the quadrupole window together with the TIM scan. So that looks like this, and as you can see, we now address, let's say, the peptides on this red bar and because it's like diagonal in the, in the heat map. I named it here, and I think we tend to name it diagonal DRA. And yeah, if it continues, the quadrupole jumps back, and then the TIM scan, second TIM scan is gone, and so on. So we address the blue area, and if you, if you have noticed, the area of peptide content is actually namely the same than on the the former DRA scheme with the quadrupole jumps, but we address the same area with less TIM scans. So in this case, we just have, in this example let's say, we just have 6 MS-MS TIM scans. I always add a survey scan in MS. So we reduce this cycle time to 700 milliseconds from 900 to 700, which is again a plus of 30% in duty cycle. And it's like now 4 times faster than the selected mass DAA alone. And yeah, with this additional duty cycle, you could, for example, reduce— if you can make the chromatography faster or you reduce the window size to increase selectivity, it's all very flexible. And yeah, as I mentioned, since 2020, some collaborators in the bioinformatics community work on processing tools, making use of it. And one very prominent collaborator is Stefan Tenzer and his group from the University Clinic in Mainz. And he is proposing, I would say it's even an additional dimension. So what Stefan Tenzer recognizes or envisions— he actually filed a patent already in 2020, so right before we provided this new acquisition scheme— and he is suggesting that you can do on each— and I focus now on this chart of this figure— each fragment in the DRA has a certain pattern, yeah, like this diagonal pattern. And this pattern is highly related to the precursor. That is because the precursor is scanned by the quadrupole and the scanning quadrupole is like, say, I want to make it here really easy, the explanation. So the scanning quadrupole is introducing a pattern which is related to the precursor. So, and if you see such fragments which based by an AI algorithm and also again machine learning could be correlated to each other. So you see a high intense fragment and a very low intense sparse fragment, and you could calculate back to the precursor, to the precursor mass, to the precursor mass, and that is within the range of plus minus 2, plus minus 1 Dalton. That means we have a selectivity on this experiment like in DDA, yeah? We have a one Dalton mass isolation window just calculated out of this diagonal DRA scan. And that is really, yeah, I would say it's adding a dimension, it's adding the precursor correlation dimension, even if it's mathematically not a dimension. And with this, Stefan Tenzer can directly retrieve DTA-like spectra out of the dataset, out of the DIA dataset. So really, you no longer need to have a spectral library. You don't need any library. You can do all the applications where de novo sequencing is required. And that works pretty good. And this bioRxiv publication Stefan demonstrated that the spectra quality is much better than on a comparable DDA run. So really this will open DAA to, let's say, an additional set of applications. So that is coming, it's not yet published, the software, and actually Bruker is working to introduce this technique Maximum Information DRA or MediaPassive into the ProteoScope software. There's another very interesting publication named SynchroPassive, again from Patricia and Matthias Mangroux, and she is using the same technique, so the diagonal DRA, and she is— I put in this picture, this figure here because it's It's very interesting. So this is like DDA acquisition scheme. You have make use only on small spots, make use of the peptides with the, yeah, she's calling it here current DRA. There you have the scanning TIMS and this the quadrupole selection and then jumping here. You see this is the peptide content and if you scan along this diagonal, you see it's much more peptide content and these bars here show actually the difference. Yeah, it's 3 times higher fragment contact in this SynchroScan. Also Patricia is introducing in her group, I think from the informatics, namely Zander Willems and Georg Wallmann, they also using kind of precursor correlation so they can filter, let's say, the resulting fragments and increase the probability for the library matching. It's still, I would say, more a classical library matching approach that will not generate DDA-like spectra, but also very interesting thing. And yeah, here I want to really encourage you to look for upcoming publications or releases for Alpha-DAE, that is the tool Matthias Mann group is developing, for Spectronaut, so Biognosis is also working on an updated Spectronaut version which can make use of the Diagonal DAE and it shows already a prototype, showed very promising data. With faster cycle times and increased ID. That will be for sure released around ASMS. So I think it's a good practice of Biognosis to release Spectronaut just in front of ASMS, and that is the plan here to then also provide a tool for this Diagnode DA and the other broker Proteoscape. The release date is not clear yet, but it's also, I think, to come in the next couple of months. So with that, I end the part of the DAA. And yeah, now we go to the future of, yeah, and current R&D work at Bruker. And yeah, I have named here, this is how how AI like OpenAI ChatGPT would illustrate the Timmstorf. And then I tried another illustration of a future Timmstorf that looks like this. I think the AI is not yet there, but those of you who have maybe seen recently the announcement of OpenAI Sora, which can generate videos out of of text input. That's really kind of fascinating and terrifying at the same time, but maybe the future of mass spectrometry development will also go in that direction. Who knows? At the moment, I cannot imagine. You see, it's more like funny than real suggestions here. But anyway, what is coming from the Bruker hardware groups, and I just want to acknowledge that we have meanwhile hardware development, or we all had all the time, also in Billerica where Matt Park is heading the R&D group, and so in a little close to Boston. We have the site in, in Vellanten now, and we have a pretty new site in Essen. So we also globalize kind of the R&D work. And one very interesting approach I will show you now, actually, it goes back to a patent also from Mel Park, Mark Ridgeway, and Desmond Kaplan from 2011. So the idea doing this is not that new, but meanwhile we have first prototypes running and it really looks promising in my opinion. Point of view. And this is a heat map I've shown with this, let's say, symbolic TIM scan in 100 milliseconds. And this time we operate it differently. We just would apply electrical field by here, so a pseudopotential— no, not a pseudopotential, an electrical field strength. And if you would do it at a certain voltage, then on Then what would happen, the low mobility species which are here in the upper region of the heat map, they will pass the TIMS analyser and the higher mobility species are trapped. So we have a first TIMS mobility low pass, so low mobility species, higher mass species will be passed and And here at this position, let's say, we could add a second TIMS mobility low pass with a slightly different voltage. And doing this again, the low mobility species will pass and in this case will be discarded. And what will happen is that we trap here a gas phase fractioning. So we, what we have designed here is kind of a gas phase selection. And with this, we could, for example, pass it to a third TIMS stage and do a mobility scan here, for example, yeah. But this is not the envisioned use case. I will show you now based on this principle. So I will call this now, these first two TIMS mobility low passes, I will call TIMS mobility filter and the third TIMS cell, I will call the TIMS Mobility Analyzer. Having this definition, I can show you some new workflows we want to implement. And so we have now the filter here, remember we can select a fraction of species out of the heat map, and we have the analyzer here. And what we propose to do is then doing DIA passive with fractioning. In this case, we would discard the species here and then transfer once one selection or one fraction, and then we would apply the standard diagonal DRA TIM scan, so quadrupole isolation, and we would scan here. And if the quadrupole reaches that position, we would step by step pass the other fraction seamlessly. So the TIM scan will remain the same. So the good thing is data processing will also remain the same. So in this case, we have the processing software right away, and we have another capability here, and that is parallel accumulation. If you remember my picture of the TIMS filter built out of two TIMS cells, the second one acts as an accumulator. So we accumulate always in second, and we scan in the third and this works like this then. So we do accumulation here, transfer, scan and accumulate the next segment, accumulate the next segment and once we are at the end of the first diagonal scan, we have accumulated already the first fraction. So that means we can seamlessly accumulate in parallel to the sequential fragmentation, which then gives us again high duty cycle. And yeah, I think I just repeat the animation here to see the accumulation on the left and the scanning on the right. And yeah, so what are the benefits of this mode? So what we could do is we could convert the high sensitivity of the TIMSTOF Ultra to higher dynamic range and deeper proteome coverage. So, from the timsTOF Ultra, we can make use of this technique already for sample amounts above 20 nanograms. On 20 nanograms or lower, so the single cell application, it would not make sense to apply this technique. But if you apply like a microgram of sample, if you have enough sample amount, then you can make use of this technique and you just, in the best case, you automatically adapt the size of each fraction to the sample amount, yeah, in like a, not an automated gain control, but like an ion charge control realised by this fractioning in the first TIMS filter. So this is something promising We will do some proof of concept measurements, had already been done, but we don't have yet the acquisition software to run everything over an LC run, so we only can do proof of concept experiments at the moment. But another very promising operation of the TIMS filter is what I called here PRM, and PRM passive is already possible with the TIMSTOF, but we would add a special mode for low-abundance species. So, for example, if you're in a targeted approach, know that here is a peptide of one protein of interest with very low abundance, which we not can quantify on a standard PAM experiment, then what we could do is that we discard the upper and the lower region and actually operating the TIMS filter with static voltages, what I have shown, will increase the mobility resolution of the TIMS, yeah. Scanning very slowly increases the mobility resolution compared to fast TIMS scan and having it at static condition gives you highest mobility. So at least 200 mobility resolution should be possible maybe even higher. So Francisco Fernández Lima showed resolutions up to 450 on a modified TIMS analyzer. But yeah, we are not yet there to spec this, but we could have this narrow isolation and then we would accumulate, yeah, we accumulate just this fraction and we can accumulate over a pretty long time. So for the TIMS-OFF Ultra on, for example, plasma, we run with 5 milliseconds accumulation time. And here you could run with like 100 milliseconds, which is still fast, but you would have a factor of 20 increase in sensitivity and protein depth coverage for selected targets. And yeah, if you have this species now accumulated, we can apply the quadrupole filter And as it is targeted, we can select the window size. We could go down to 1 Thomson if necessary, if you want to have the isotopic patterns, and we make it a little bit broader. But this really, I at least, I'm pretty sure that this will give very good PRM performance on the TIMS TOF and Yeah, you see we will increase the mobility resolution at least by a factor of 4, so higher selectivity. We can combine it with a narrow quadrupole window, and we could even combine such PRM targets with DRA, diagonal DRA. But I think a general PRM approach where we typically run like 80 targets in a classical, let's say, PRM where we address a lot of targets in a standard TIMS scan, and then we could add like 3 high sensitivity PM targets with making use of the mobility filter, TIMS mobility filter, and this will give us 1-second cycle time. So let's say 83 PM targets plus an MS per second, that should be easily possible. And of course you can have it flexible for medium abundant peptides, you can reduce it and then address more. And so it will be flexible, that's for sure, especially on PRM, you have all the possibilities. And finally, I want to show you a last usage or a third usage of the TIMSS mobility filter. And that is excluding high-abundant species. Like the high-intense albumin peptides from neat plasma. And what we would do is we do a survey scan, so the MS scan, and have a detection of the high mobility, of the high abundant species, and we would dynamically filter them out. So in this case, yeah, it works like this. You can imagine we scan in diagonal DRA the first fraction of the heat map, then we, yeah, stop here, let's say, and then we could decide now the high-intense precursor is eluting. We could either skip it, so jump above this mobility region, or we scan it but with a very short accumulation time. So the filter has normal accumulation time here in this region, like 5 milliseconds, and then we have even lower or no accumulation time on the higher abundance species. Here it's still in the heat map, and then we scan with another short accumulation time the rest of the heat map. And yeah, with this we have, yeah, first time in my presentation now at the end, we have a slight disadvantage induced in the cycle because we need two accumulation times. So if we If you have one precursor which we want to deselect, yeah, filter out, then we need to accumulate 2 times. And if you want to exclude, like in this example, 5, 4 precursors, we need to accumulate 5 times depending on the distribution of the precursor. We can do parallel accumulation, but if they are distributed not evenly spaced, then we will lose in worst case for each excluded precursor the echo time and doing some math here for this example, if you want to exclude 4 targets, we lose about 80% of duty cycle, which is still great. And further than that, Aaron Pansek from our research group in Billerica together with Ben Jones and Mark Ridgway, they have already shown that they see a huge increase in identified proteins out of neat plasma with this technique. Yeah, and with this, I'm at the end of my presentation, hopefully kind of in time. And yeah, I exceeded a little bit. And I would end with the acknowledgements. And I'm happy to answer questions.

Fantastic. So thanks very much, Oliver, for this, this very deep dive on the, on the technical aspects of TIMS. We do have a few questions. I can, I don't know if you can see the Q&A, or I can try to summarize for you. So the first one concerns The myDIA passive, whether— so you mentioned Proteoscape software. Do you think there will be other software solutions available, in particular open-source software solutions like Skyline? Are there any activities on this side planned?

Actually, I do not know about Skyline. I hope that the Vadim Dimitrov group is also working on processing software maybe also for the media. I mentioned that that is a patent of the University of Mainz, so I think that has to be negotiated with Stefan Tenzer or the university to get a license. I'm not 100% sure how this happens, but I really hope that— so the diagonal DRA I showed is already since ASMS last year included in the acquisition software. It's called VistaScan there. And I think that users will acquire data and all the bioinformatics groups can acquire data if they have access to a timsTOF or a collaborator. So I hope that this will take pace and that all other tools and vendors of processing software will jump on this. Maybe they need to be convinced by the protein performance in the end, yeah.

Yeah, okay, good. So, and then the second question is also about the, the sort of diagonal, uh, schemes. Whether does Oliver think that, that, um, these should have utility in small molecule applications, kind of similar to, to PASEF, which was originally, I guess, developed for peptides but then, uh, implemented for lipids and so on? Do you see such thing happening for the diagonal schemes?

Yeah, definitely. So I, I knew that our applications group found already its advantages to use it for lipidomics. And especially the media or myDIA, how you termed it, that is capable to do, yeah, spectra, DDA spectra retrieval without spectral library. So you can do any application with that and still get DDA-like spectra out of a DIA run. So that should work. Not sure if the software supports it. So really the software is, maybe you have recognized that Stefan Tenzer and Bruker started 2020 to implement it and it's maybe now close for release, but it's high effort, I would say, and it's just the group of developers is not small there.

Yeah, yeah. The next question is also about small molecules and especially less than 150 M/Z. Is there any development that might improve analysis for these small molecules using TIMS, like the small small?

Yeah, actually we are working on improvements here, but it's too early to give any concrete statement. We know about this and that you actually lose duty cycle. So we have this, I think we call it TIMS stepping, where you have a very small molecule scan and then another scan on higher masses, but then you lose 50% of the duty cycle. And for most applications, I think for small molecules it's used from time to time, but yeah, we know about this. And yeah, but I cannot announce any concrete result or development here.

Okay, very good. Fourth question relates to library Matching from DDA to DIA analysis, how important is it to have the same LC conditions? Is there some flexibility there?

Yeah, I'm not expert on this, maybe rather you, Ben, but I would say the DDA generation of a library is no longer that important. So the the colleagues from Biognosis, they mentioned that you maybe get a low single-digit percentage increase if you take the effort and generate on the same system with the same LC a DDA library. But the predicted library, or I think it's called Direct DDA at least in one tool, that is most of the time used and for everyday work, especially if you run DDA across different instruments maybe even from different vendors, then the direct DAA approach is much more advanced and easier to integrate.

I agree with this. I mean, I think you can still top up a little bit with these larger libraries. And in terms of the chromatography matching, many of the softwares have now nonlinear alignment in the retention time space. And I think this makes a big difference if you want to match across LC conditions that are somewhat different, although the further you get away, the less good it is. I would put it like this. But I kind of agree, Oliver, that it goes in the direction of direct analysis. Okay, do you think it's possible to create a type of spectral library for the 14 more abundant proteins using the dynamic mobility filter? And he, this person, even proposed a name, digital depletion. He wants to call David Gomez Varela.

So I actually, maybe I do not get this fully. So there are 14 abundant proteins.

I guess he was talking about the plasma data that you showed towards the end there.

Yeah, I excluded 4 of them. I think identifying them is not that much of an issue.

Yeah.

And they will be low abundant species on the tailing. Of the chromatography. So typically, they tail a lot. And I think we identify them all the time. Problem is that they suppress other ions, other low-abundance species with the same mass and mobility. And here, you really— the deselection worked pretty well on a few of the albumin peptides. This is maybe kind of what I just mentioned at the beginning, Florian Meyer and colleagues did program it manually. That is the status now. Yeah, so really they programmed manually to exclude the high-abundant albumin peptide species and showed great progress. But now all the software acquisition software needs to come together and do it then in an automatic manner.

Yeah, okay. David, I don't know if we answered the question. You can write a clarification if you like, but I think I think maybe. And then, so the last one— people keep adding questions, so I say the last one again. So with respect to the fractionation, how do you keep track of calibration in the IMS dimension? Does it become more challenging, requiring more evenly spaced calibration solutions that would show up in all fractions? So how do you calibrate with this fractionation in ion mobility dimension?

Maybe just, just one comment to the former question. I like the digital depletion It's a good name, yeah. So actually, calibration is not affected for two of the three things I propose. If you have this third TIMS analyzer where you do the fractioning, let's say, very systematically and transfer the fraction right in time for the analyzer scan, then you just keep the calibration of the TIMS analyzers. If of course we need a calibration of the TIMS filter, that is anyway needed. So the static calibration of the TIMS filter, that is an additional calibration we have to implement. That is possible, but it's not yet seamlessly implemented, one of the tasks we have to do, of course. And the same is for the PRM approach I have shown. There you just need to calibrate the filter and then you have the mobility of that special target species. The classical or the higher abundant PRM targets will be run as normal and then we add some species. I had a nice conversation with Tejas, one of, I think, the R&D manager of Biognosis, and he said it's not any problem to to include those targets into their PRM software spectrum line. So that will be very easy. So I guess we get it right away when we have the acquisition scheme ready. And yeah, I guess that will then also be the case for the other bioinformatics platforms. So calibration, there is, I would say, it's some work for the filter calibration, but it's not challenging or not risky or what, yes, I think it's just work to be done. And for the third mode, there it might happen that calibration is more challenging because we need the service scan directly giving you the same mobilities as then the diffractions and stopping and starting the TIM scan, for example, if we need to, or jumping across What we did so far is that we calibrated both parts of the scan separately, and actually it's one and the same calibration. If you have, say, the same scan speed on the TIMS, then the calibration can easily be transferred. Only if you change, let's say, the slope of the speed, then you need, as of today, recalibrate. Christian Bleiholder published an approach where you can have how is he calling this paper, is about unified calibration. And then he, with a set of experiments, he could calibrate the TIMS analyzer independent from the slope. And so that might be some approach, but yeah, that's definitely a question which needs an answer.

Okay, and just maybe one further question from me on that is because the hardware, You need serial TIMS analyzer, I guess, to make this work, as in, are we likely to see new hardware coming to support this at some point, or what's your feeling about that?

Yeah, so at least the TIMS filter would fit in the TIMS TOF Ultra or in the TIMS TOF SE. It would fit in any TIMS TOF.

So it could be an upgrade, you mean?

Yeah, so that could be an upgrade. I would say it makes only sense for the TIMS TOF Ultra because you need the high sensitivity in order to to have the short accumulation times. If you would do it on, let's say, in TIMSOF HD or older TIMSOF, then you would lose duty cycle by that, or less gain, let's say, in filtering. So the PRM workflow definitely could be an upgrade. I don't know if that is decided to be an upgrade or if a new product will be released. We have, of course, a lot of other components in development. So maybe we release a new product and combining a couple of components, but yeah, in the past we always offered also an upgrade if possible. For the other two scan modes, we need a third TIMS analyzer. So TIMS theoretically, it fits in the Ultra because we have this single cell cartridge that we could integrate another TIMS but that's not the best performing design, I would say. So possibly we will come up with a new design on this TIMS filter.

Okay, great.

Yeah, but okay, nothing I can stay for granted. Yeah, it's not yet decided, it's still in development, so Good.

Okay, well then I think we can start to finish up because it's getting a bit late, but then we finish with maybe one controversial question, which you can answer or not, but some in the chat would asking questions about comparisons between TIMSS TOF-Ultra and for example, ASTREL. And there's also a proposal to have a debate at ASMS between the, some kind of face-off between these two. So do you have any general comment about the comparison between these two types of instrument?

Yeah, of course, we look closely what the R&D guys at the other side of the River Weser are doing. So they are also located here close by in Bremen, that's what I mean. And yeah, it's a very interesting instrument, of course. And I would say it gives surprisingly, from the publication I see and from users, it gets surprisingly good results. Actually, I see some compromises on that system, but as it performs good, it's a good combination. And yeah, really, I think for the community it's good if there are as many as possible different hardware configurations competing and that is also for us a nice challenge to have.

Okay, good answer. I think we can wrap up there. So Giuseppe, did you want to say something?

Yeah, I would like to thank Oliver. Thanks, Oliver, for actually this great overview on the history of Teams and also showing us all the different flavors of the Teams that have been developed. Also all the participants, there's been a lot of questions actually. Maybe I might ask just last one, Ben, just before. Sure. So, Oliver, I was thinking if you could actually tell us more about some application of the TIMS going towards intact proteins or even complexes.

Do you have— Maybe you have heard about, I think we had a team from Essence around Dimitris Papandreou at the ASMS in our group that is a former Phasmathek or we call it now Bruker Phasmathek Group. And they developed the so-called Omitrap. And we are, I think it's no secret, we are working on integrating the Omitrap into the TimsTOF. And that will give us great new fragmentation techniques, all what we call ExD. Plus UVPD and other things which are possible in the OmniTrap. So that will be, I think, really a great addition for the intact protein community. Yeah, but we are also working on improvement. We have an upgrade for the timsTOF Ultra where you get way improved dissolution, which we currently test with collaborators. So Yeah, there's just also a lot going on on the intact protein complex or antibody application side.

Thanks a lot. Ben?

Fantastic. Yeah, thanks very much, Oliver. Thanks to all the participants. So I think we can close it there. I think the next webinar will be in March, Giuseppe, right? And it will be— the topic will be reverse phase protein arrays, if I'm not mistaken. So watch out for the announcement for that. Fantastic. Thank you very much. Then we'll end it here. And thanks again to everyone. Bye-bye.

Yeah. Many thanks to the audience for attending that long.

Thank you.

Yep. See you soon. Bye-bye.