 Welcome everybody to the Biaxel webinar number 59. Today we have X3 DNA or DSSR, a resource for structural bioinformatics of nucleic acid. The presenter of today is Sean Chu Lu from Columbia University. I'm hosting this webinar together with Stefan Farr from University of Edinburgh. Today presenter is Sean Chu Lu, he's a scientific researcher at Columbia University, he's a specialist in nucleic acid structural bioinformatics, and he has a huge amount of publication connected to 3DNA and DSS software. So now I will give the word to him so he can tell us about these tools. Hi everyone, I'm Sean Chu Lu, I'm the developer of 3DNA and DSSR. Today I'm very pleased to share with you X3 DNA DSSR, a resource for structural bioinformatics of nucleic acid. DSSR stands for dissecting the spatial structure of RNA. It excels in structural bioinformatics of nucleic acid, including DNA, RNA, and their complexes with proteins. DSSR performs structural analysis. It is set along with some other well-known software tools by leading scientists in top journals. DSSR produced block wheel schematics with unprecedented clarity. DSSR has advanced model building capabilities. DSSR can be easily integrated into other bioinformatic resources. DSSR has unique features for g-chordoplexes, under-promise over-deliver. That is the principle I have followed while developing DSSR. Published results and documented features of DSSR are reproducible, period. In fact, DSSR has a lot more powerful. It enables innovative cutting edge applications. To better understand DSSR, we need to go back in time. DSSR is built upon 3DNA. 3DNA stands for three-dimensional nucleic acid. It was developed around the year 2000, while I was a post-doc at Wima Austin Lab at Rutgers. Here are the two key publications of 3DNA. One is 2003 NR paper, another is 2008 Nature Protocol paper. These two papers are well-set. Even today, 3DNA NR paper is still set at roughly 100 times per year. More recently, we have built an updated web interface for 3DNA. And this is the result you will need to get access to the new features. Published work was highlighted in new class research in 2009 web server issue. If you are interested, the color image itself is reproducible. 3DNA calculates DNA shape parameters. I think some of you are familiar with the figure on the right. This figure was initially created in the 2003 NR paper. Over the years, it has become very popular. To my knowledge, it has already been adapted into several textbooks, including introduction to protein DNA interaction by Gary Stommel. The 3DNA parameters have been adapted into nucleic acid database, NED. Here is one example of a DNA structure, the famous Dixon-Dickerman. Here is the citation of 3DNA of the two 3DNA papers. 3DNA also laid a foundation for 3DDART and 2x3DNA resources. 3DDART stands for 3DNA-driven DNA analysis and rebuilding tools. It uses 3DNA rebuilding functionality. 2x3DNA extends the capability of 3DNA package to grow max MD trajectory analysis. DSSR is the next generation of 3DNA. I have continuously developed DSSR for more than 10 years, taking advantage of my expert domain knowledge of nucleic acid structures, detailed oriented software engineering skills, and extensive user support experience. DSSR includes all the battle-tested features of 3DNA, plus many, many more. The DSSR logo has X-ray DNA in it. So far, I have published 3 DSSR papers all in nucleic acid research. The 2015 NL paper focused on the structure analysis and annotation of RNA structures. The DSSR demo and the DSSR panel integration were published in 2017 and 2020 respectively. DSSR has three major functions, identification and analysis, block wheel schematic, advanced model building. It can be integrated into other resources very easily. It has a unique feature. In the following, I will give a talk on each of these topics. First, let's focus on identification and analysis. I will use some typical examples built on literature citations. Before I do that, I will give you a quick overview of how DSSR actually works. DSSR is a command line program. All the features I'm going to talk about are integrated into a small, executable X-ray DNA DSSR, which is only 1.8 MB. It has no external dependencies so that you can easily get started just starting with a main edge for health. DSSR can take PDB at the input or MMC at the input. It can also generate a JSON output for ED processing. For the classic tRNA, you can run the program like this. You give the input and you give the output. The program automatically identifies 7 to 6 nucleotides, including 40 modified nucleotides. It identifies 34 bit pairs. Two helices carry bound the two arms of the arrow-shaped tertiary structure. Four stems. It identifies three happy loops, including the anti-coding loop, and one junction, which is roundabout in the 2D clover-leaf secondary structure. You get an output of the bit pair, where you can see it gives the identity of the nucleotide form in the pair. Also, I can give you the name of the bit pair, including wasn't correct bit pair wall pair and no wasn't correct bit pair. You get a reward hook thing, a reward wasn't correct bit pair, and also classify the bit pair using some annotation, Leontes-Westhoff annotation, and one of DSSR unique annotation. Moreover, DSSR generates secondary structure in three common formats. This file can be fed directly into Warner to generate this 2D diagram you are so familiar with. With that background, let's see what other users are making use of DSSR. The Raman collection line, use the DSSR to extract bit pair information of a Raman sum RNA in the publication in science in 2017. The D-shell research team use the DSSR to analyze bit-stacking interaction in single-strand RNA tetramer in their RNA force field. In the Sijie Chen lab, using DSSR to identify 2D structures from 3D atomic coordinates, DSSR not only automated the process, but removed human errors. It saves your time, it gets the job done. To nature paper in 2020, DSSR have been used to identify wasn't correct and no wasn't correct bit pairs, and have been used to calculate shape parameters, including growth width, helical parameters. Along with time, DSSR produce block wheel schematics that are simple, effective, and acytically pleasing. In block wheel schematics, DSSR generate blocks, bit blocks that make the bit identity, bit pairings, and stacking obvious. Using this color code, following the NDP convention, G8 green, C8 color yellow, A right, U8 cyan, and T8 blue, you can immediately see in this figure, there are two stacked eighth color right. The one mismatch, GG pair, and the GC pair, green and the yellow, and the U8 pair, cyan and a right. You can also immediately see the board of the U, colored cyan. On top of that, DSSR produce wasn't create bit pair block that reveal the double helical regions and the growth. The main growth is colored black, the major growth is water color base is associated with. You can immediately see this main growth, this major growth, and this main growth is in you, and this main growth is in you. Moreover, DSSR can automatically identify G-Tetris, and they use a square block to reprinted this G-Tetris. Using this representation, you can immediately see that this structure contains two G-Tetris. You can also see the looped residue, this is the G, colored green, it's a T, blue, and the C, yellow. To give you a better understanding of how DSSR's timeout schematic compared with other, let's have a look of the PEPP term in complex way as we see a recent publication in nature chemical biology. On the left is a published result. On the right, it generates automatically with DSSR. You give your couple of seconds to see for yourself. As you can see the DSSR can give you a major growth very clearly. This is the double collection, which major growth is in you, this is the helical loop. This is the stacking of C, A, A, U. Now this is as DC, the second, and this is a quadruple plate with U, with C in yellow, U in cyan and G underneath it. This is very clear in this schematic, where it may not be that obvious from the left one. Also this generates automatically in less than the second. I don't know how much effort you also spend to generate this figure. This is another example of a chair type telomeric DNA G-chordoplex. Here on the left is taken directly from publication on the right is DSSR, PEMO generation. Again, you can see the three-layered G-chordoplex. On the bottom of it, simply by color coding, you see the A, A mismatch pair, which has a propeller on it. This is not planar, that is propeller, large propeller with this bit pair. The block-wheel schematics are highly acclaimed. In the factory opinions website, DSSR website has been rated as very good. It is classified as good for teaching. The web interface was recognized as very simple and effective and highly recommended. Here are the 12 power images of our journal in 2021. The images were provided by the NDB. They were generated using DSSR and PEMO. The 2020 NAR paper was set explicitly. This is a single author paper I have published in my scientific career. You can generate all these schematic images very, very easily by using this web interface. Here you can just click a PDF ID or you can upload it to your structure, simply by using a URL or upload any of your PDF or MIM file. You can customize it using some common options, or you can download the PEMO session file to customize any way you like. DSSR also has advanced model building capabilities, including RNA modeling, DNA protein complex, G-chordoplex. In addition to 3DNAs, you know, fiber model, customer model, base sequence, and the parameters. 3DNA DSSR has been used by MOOC scientists for RNA modeling. In April 2015, I received a message from MOOC scientists asking for permission to use a modeling software I have developed for their project on drug design. I was so happy that 3DNA could find a use in a big farmer and I approved their request. Later that year, I noticed this nature publication by the MOOC scientists. In this paper, 3DNA was explicitly stated that the homology model would contract muting program mutated bits of the 3DNA package. Notice that in 2015, the DSSR paper was not published yet. This special module was distributed as part of 3DNA. In a follow-up paper by MOOC scientists in 2017, 3DNA was set again in this way. It was noted that this module had been integrated into DSSR and substantially improved. DSSR can perform template-based assembly of DNA protein complex. Here are two chromatin-like models. They are built using PDB entry 4xZKL. By simply varying the length of linker, you can see that it can generate two quite different topologies of this model. On the left, it's sort of like stacking. On the right, it's more complicated. I generated it simply for fun because the program can do it. I'm sure there are some applications for this module. DSSR also can generate a circular DNA with perfect plan geometry. On the left, it's a schematic wheel of 100 bit pairs. On the right, it has 150 bit pairs. These are generated in the command line in less than a second. DSSR also has unique model building capabilities for G-cordoplex. Here it's just like a schematic of a ladder, a rack, which has no twist. All the G-cordoplex have a similar face. On the right, I introduced a 30 degree twist and also alternating the faces of the G-cordoplex added up or down, color green or black. DSSR itself can do a lot. It can achieve much, much more by easily integrating into other resources. The DSSR-GEMO integration brings the secret like queries into GEMO. For example, you can select junctions or select no walls and crack bit pairs using the new GEMO command color in right. This work was done in collaboration with Bob Hansen, the principal PEMO developer. This work was published in ANR in 2017 and eight words highlighted in the cover. Again, if you are interested, the color image is reproducible. The DSSR-PEMO integration brings innovative schematics in new collect head structures, not only at the bit pair level, but also at the bit pair level, at the G-cordoplex level. This work was done with Thomas Holder when he was the principal PEMO developer. Thomas wrote the DSSR-PEMO plugin so that users can generate this schematic interactively within PEMO. In addition to DSSR-GEMO and DSSR-PEMO integration, which I initiated, DSSR has also been used by so many other bioinformatics resources. Here is a list of the 16 published ones that cover a broad range of topics in structural bioinformatics of new classes. For example, the URS stands for unit worth of RNA structures. Our type of product is for G-cordoplexes. DNA-proDB stands for DNA protein database. And the list keeps growing. PDB redo also makes use of DSSR as you can see from our logo. And I'm honored to be a co-author of a recent publication with PDB redo, which provides a new method for restraints and validations of new collect head structures. DSSR's systemic integrated approach enables no application to be developed quickly. Here are two examples I have developed over the past couple of years. One is the trans-infact DNA complex containing 5-method C. I perform analysis of all these structures in the protein database. And this work gave you a rationale of 5-method C recognition by trans-infactors. And I also created, I noted the list of the integrated motif, so-called I-motive in the protein database, which have the C plus C path. This is an unpublished result, but in the following, I will focus on one thing I have done, which is already available on the website everyone can have access. This is a feature for G-cordoplex. I have already talked about the schematic and some modern capabilities. In the next, I will focus on just on identification and annotation of G-cordoplex. So a very simple question, how many G-cordoplexes are in the PDP? You would assume this is a very, very simple question, but the leading authors can come up with very different answers. For example, in the 2021 paper by Stephen Needle, a given number of 520 in another publication, a given number of 246, they differ by a fact or two. So which one is correct? It turns out they are all off the market. The actual number at the end of 2020 is 372. As of early this month, the number is 450. How do I know? Because I use the DSHR to identify the G-cordoplex automatically from atomic coordinates. A given number without a dot. Just to give you a quick overview of G-cordoplex. Uniform unimolecular G-cordoplex are formed by G-tetris connected by the loop region. The loop can be lateral or diagonal or propeller with a double chain reversal. Depending on the order of the loops, the G-cordoplex with a false trend can be any parallel or mixed. Overall, the G-cordoplexes are highly polymorphic, much more complicated than the double-flex DNA algorithm. In 2018, the Weber-Dieselwer lab published a new structure detractor to characterize G-cordoplex systemically. Here is figure one from the figure, giving the definition and illustrate with a sixth example. For example, for 140d. So here the two means there are two layers of G-tetris. The arrow means a lateral loop, and N and W means a narrow and a wide growth respectively. However, in the figure one of the six examples, two of them are described as signed incorrectly. For PCB 2DKU and 2ROD, the growth width N, narrow, W, wide are reversed for three cases. And I communicate with Weber-Dieselwer and he acknowledges the mistake. And currently we are working collaboratively on the reverse version of the descriptor. In DSSR, I have accumulated a list of all the G-cordoplex structures in the protein data bag. The result is called G4DB. And the results are presented using a dynamic ITML table, which allows flexible searching. I know that these have been used by some experiment list in the field. For example, simply typing the optimal in the search box, you will immediately feel that there is 64 entries out of the 415 structure. If you combine optimal and the chair conformation of the structure, you will immediately reduce the heat to 22. For each of the G-cordoplex, the DSSR G4DB provides a comprehensive annotation. Here is a small portion of it. Where you can see that Weber-Dieselwer normal culture is automatically derived. And also program theory or a common name, this is a chair type of structure. The DSSR also calculate a rigid body parameter here is right in the twist, using the same formula as for DNA double helix. The DSSR also gives a detailed list of the loop given the type and identity. From here, you can make use to get all the information you need for follow-up analysis if you want to do. So far, I have showed you the most visible features of DSSR, including identifying analysis, block wheel schematics and model building. And also it integrates into other bioinformatics resources and special features for G-cordoplexes. Under the same, however, that's a lot of hard work have been done. I have the details are essential for software product. I have developed a DSSR using strict ASIC and I treat warnings at errors. I use the most stringent compiler options that have available to test the program. I also use well-grant to check for memory leaks so that the program can run without memory leaks. For each major release of DSSR, I test the program using all the new class structures in the protein data bank so that I can be sure that the program works. So in the end, I will say DSSR is an integrated software tool with unmatched capabilities in RNA and DNA structure bioinformatics. Here are the three major web resources you can make use of. They are all available on the X-rayDNA.org. First with Web X-rayDNA.org which covers all the new features of X-rayDNA in a very easy to use web interface. The schematic dot X-rayDNA service is not only provides the schematics, it also provides JSON and a human readable output so that if you want to get a feeling of what DSSR has to offer, just go to this website. If our interest in key quadruplex in the protein data bank, this is a resource that is well worth your time to check out. DSSR is a command-line program. It is very, very small. It's less than two megabytes. It has no dependency of any third-party library whatsoever. It can be set up. It has no setup, no configuration needed. You can get it up and running within a couple of minutes after you get it. If you want some example, this main edge or main curve command option, you'll get it started real quickly. Also, there is a professional user mining currently at 236 pages. And finally, I would like to acknowledge Hama Boothmarker, William Olson, and Brad and the best for our health. And the 3DNA user community. Thank you for attending the webinar. Let me show you one thing I want to make sure. Since we still have time, I think I can give you some left demo so that you can see how DSSR really works. Okay. Well, here is a folder with these three files. Just type X3DNA DSSR, main edge. You should get this information. This example, I'll show you how to get started very, very quickly. For example, if you get X3DNA main edge, one EHD, PDD, you will get this output very quickly. It takes less than zero seconds. Okay, less than one second because it's too little. This is the output. What's the output look like? This is a bit pair. Here, this is a multiplate where it's a higher level base association in the core plan geometry. Here, just like a triplet. This is a double-collect region of this tRNA. It has two helices. And this is four stems. This is the stacking of the nucleotide, which are not in the double-collect region. Here is the list of the three helping loops. And this is the junking loop. In the end, you can see that several additional files where you can have access. If you have a list, now you can see there are many more files here. This .dbn.ct are the second structure where you can easily use to generate a 2D diagram. Let me give you an example of how it works. This is one of my favorite 2D structure viewer. Let's delete this one. And then open this file. Go to this demo. You can use .ct, .dbn, .bsql, .propha, .dt, .ct connect table because it has more information. From here, you can see, you can generate this figure automatically. Oh, sorry. And you can customize this in whatever way you like. There are so many options. This is very handy too. For example, you can just like to show this numbering how many bit pairs. And you can have the view. You can draw using a linear diagram. Again, there are so many features you can play with. In addition to using, let's clean it up because there are generally so many things. Now you have this one. You can also run directly using a same file. Again, the program run in a similar way. Moreover, you can use the JSON output so that you can generate the figure, the output in the structure format that can be easily integrated into some other resources. The JSON output is one line. It is not for human to read. But if you have some other GQ, it's one of the passers, command line passers of JSON. You can see, this is how you can generate. This is a junction. This is a four-way junction where they give you detailed information. For example, you can use the pass and give you all the bit pair information. So this is very, very easy to use. Let's give you another example of here is a schematic. This is a website of schematicaccidently.org. Here you can just on the top. You can just go or give you some quick overview of how they struck. Here is a panel structure schematic look like. If you like, you can download some of these figures. From here, you can input for all the new class structure in your predicted bank. They are already calculated. I've updated this list once every week. For example, if you put one EHD, you will get this output. Here is just like a very simple overview of it with some meta information. And also with this schematic in sixth orientation. I will draw your attention to this output. This is a text output and also that JSON output. If you want to just have a general idea of what schematic look like, you can have a quickly, you know, this is a sample selection, random selection of the 30 or 60 inches. Oh, this is a good example. You can see that there are a lot of A's. Okay, this is A, color right. Five of them, four layers. On top of it, it's just like a five T's. So sometimes, you know, you can have a quick look of this. This must be ADNA, mangrove, magic groove. This is DNA protein complex. So let's give you some general idea on how you can, in this part, you can specify, you know, the PDB or MF6 coordinates. You can use, you know, the PDB file or MF6 file, or you can upload a PDB file using this option. Let's use the, you know, the MF6 and just run the program. Give you a left demo to see how it works. Maybe it is running. Okay, so this is the result. You generate it automatically. You can download this, all this image in the table file. You can download the panel session file, where you can play around to do customized in the way you like. At the end of it, you can also have the structure feature in text format, injection format. Let's look at the G4. This is the G4 updated early this month. You can use, you know, F-T-A-M-E-R, see? That's 64. You can just go to your phone. You go to 22. This is what I used for the slide. If you want to have only accurate structure, so you have only 17 left. So you can sort this in a very simple way. When you click on this, you will generate annotation. Now, this will give you an example of what annotation looks like. Here, just like for the helices. And this is the stacking diagram, where you can see how each of the steps are stacked. At the top, at the bottom, you have what they call the G4 stem. It will give you the several annotations. And this is the common name. This is the detailed annotation I show in the slide. Okay. So that's what I want to share with you for now. So if you have any questions, I'll be very happy to answer them. Okay. Steven, that's up to you. Yeah, thanks for the presentation. It was very nice. And the demo was very interesting. So there's no questions in the chat right now. If anyone has questions, then please type them in. So I have a question for you. So during my PhD, as I finished recently, I was using one of your old three DNA programs. I think the fine pair program. So I think my game was just to take a PDB structure and find the rigid base pairs. So is it fine to still use that sort of old software? You mentioned you wrote quite a long time ago. So would you say is it fine to still use that software? Would you suggest that I use your newer DSSR software? And if I did, would it give me the same outputs as the old three DNA fine pair software? Okay. So I think that's a very good question for three DNA, which is called a sweetheart program. Did you encourage more than 20 C-programs plus Ruby pro script? So fine pair is one of the components prepared input for analyze. And this is just like for DSSR, all these features are integrated into one. Just use the command option. The program automatically generates the input for you and analyze it. But if you want to have some control of it, you can generate the input and they make any modification of it. And then just like a run analyze using that multi-fine list for analysis. So this just makes it so much easier to have only one program and cover everything just using different command option or different module. So it's much easier more than version of it. So I would suggest in the future, DSSR is a way to go. Also, I'm no longer supporting DSSR DNA anymore because this is for me, it's the past. Yeah, that's good. So with that, so yeah, it will still give sort of the same results, let's say. Yes, yes, you know, because there are just some subtle difference but for the base pair, for more than trick base pair, it still gives us identical numerical parameters. That's what I say, you know, the DSSR is built on 3D. All the core features are there. It's not just like a brand new, just for the type of brand new claim something new, but they build step by step. All my past 20 years, Korea, I just built one thing toward DSSR. I see I'll have to update some of our workflows, I think, to use it. Actually, you know, it's much easier to use than 3D. It got set up, you know, set up environment, variable for, you know, for the command path and for, you know, what the file look, what the look, you know, but for DSSR, everything is just one small program. It covers everything. I activated the erase hand option. So if people want to just erase their end and make directly the question, they can do it. Maybe someone has just too much to write a question. In the panel, but we have a question coming up. Yeah, so we got a question from Jiri. Is there any good reason producing B1 slash B2 annotations for RNAs? Like in the one EHZ example. Yeah, okay. So the B1, B2 confirmation, it's actually started with the DNA. And for RNA, we would normally assume it's single-trand, it's C2 prime, you know, it's, you know, but actually it's more complicated. And there's some, for example, for some of RNA tractors confirmation, it can have a C2 sugar confirmation for backbone. So basically, we won't assume B1, B2, you know, don't have it at all. But this program, when you 3D DSR analysis structure, it's big DNA and RNA simultaneously. It's the same, you know, it do no special treatment. It just uses the same algorithm and to classify whether it's B2 or B1 using either theta, the torsion angle. So that's nothing technically, you know, complicated with that. It's just using the same classification for RNA and applied to the DNA structure. Whether, you know, it makes sense to end the user, it's up to them to decide it. But given that choice, you may notice some of the surprise, you won't affect. So yeah, so the follow-up to that was, is it just a wrong misleading result? Okay, the true is the true, you know, whether it's misleading or not, you know, it's just like up to the user to decide it. For example, what several long time ago, the people had a question about the DNA protein cap complex, assuming the DNA is in B form. And just using the DNA, using the P parameter, I didn't find part of it in A form. And using this tool, analyze it consistently, maybe we will some surprise. And that part of DNA confirmation in DNA complex, DNA protein complex, this is the paper I think it's well recognized. And I think people should be open-minded and take a true as a true. Do not be prejudiced by whatever your privacy is, you know. And as far as DSHR concern, that backbone, total angle is a tiny, tiny piece of what DSHR is all for. If you don't mention it, I may not pay attention to it at all. And if you misuse that information, it actually have to you. And this is a tool for you to make use of. Many want to ask you, what will be your following step? In which direction will it go now? Yeah, there is, you know, that is DSHR, like I said, integrate all these features into one. So there is something, I mentioned the project for the 20-in-fact recognition of five mesocene. And also just like a written list of integrated motifs. There are so many other structure features in the part of the bank that can be easily done with DSHR, just like the infrastructure already put in there. So you have... I think in building structure, so building structure, some building option for different RNA motif. Yes, this is for analysis. This is for analysis annotation. But for the modeling part, this is mostly in head demand in the field. A lot of you know that... You can do analysis, give the structure, find the, you know, base pair, you know, wasn't correct, nor was some of this called motif. But building the structure from these components, it's a totally different story. DSHR is unique position to do that. Also, a lot of people are using DSHR in the analysis of the molecular dynamics simulation, you know. And there is some rudimentary support of it, but I think the more can be done, you know, how to better integrate DSHR into the molecular dynamics field. So which type of format are you supporting now? Country, you know, I'm just like... Coming from the molecular simulation or from other... Simulation for DSHR, it's not... It's neutral to, you know, gromax and broochon. It do not handle this, you know, you know, net format of that. It just started with, you know, with MMC or PDB format in the model and the model. And just the standard format generates, you know, rejection of full format. The rejection of full format can be passed. Because DSHR has so many structure features, it's already there. You can just, like, fill the whatever you want for downstream analysis. So this is just like... This is very different from, you know, what are anguar, gromax, specific analysis pipelines. And it's neutral to all of them. It has much more to offer because of data, studying its prices already. What that was meaning. Andrew, so you read the trajectory in the PDB format. I read in the PDB format. PDB, imported PDB format or MMC format. Sorry, could you say the second one? Because I could... It's called MMC. Macro, yeah, MMC. These are the standard format I want to, you know, follow through. For other, you know, binary, predatory, whatever called the format, I don't want to get DSHR involved. Okay, okay, so because there would be the challenge that those formats that require so forth, in some, if you have a lot of snapshots, they would require a lot of... Yeah, I want DSHR, I want DSHR neutral to all that. To all these, I don't want DSHR to be tied to this and that's in certain party, you know, we all follow the standard. We communicate through a standard protocol, just like a JSON, just like a standard text file. That's my philosophy. Yeah, yeah, so the, the process for processing the trajectory you would suggest that the user will extract the PDB or every snapshot, it will be read from DSHR and then extract the second one and will be read it. That is what the loop that will be. That is what, and then you generate, you keep the information also, for example, of time, when you have, when you read the PDB, you have time information in your output or not. Because when you have the PDB, PDB and the model and the model, for example, you have that information, you know, put it in the sequential order in your trajectory in the PDB file. So, you know, program just like the DSHR just follow whatever you give it to analyze. It's not that smart, but you do what ask to do, you know, rigorously. Also, just like, you know, a machine, you know, do X3DNA. This is a web service developed by some guy in Germany. So it's just like, this is built on top of 3DNA. Also, I will say, you know, DSHR has so much more to offer. The guy, you know, when you are, you know, we were mentioning, you know, there's a B1, B2 class, this is a tiny, tiny piece of whatever DSHR to offer. I do not even pay attention to that before this question would ask. And I'm sorry for whatever misleading that may give you, but this is a tiny, tiny piece of DSHR. It's not involved, it's not in any of my people to be missing at all. So that, you know, if you want to know, if you can give me better suggestions on how to clarify your confusion, I would love to hear. I have a question about the software development. You mentioned you used strict NCC for all of it. So having done some of that myself, it's quite a lot of work. So why have you chosen to program it that way? Okay, so I use C because C is a mature language. It is very simple, but not year old, sure. But I'm experiencing now, so that I don't see so well, so that programming is not an issue for me at all. The basic part, most important part understanding. So I use C, I use strict C, so that I can compare the program right away in Windows, in Linux, in Mac. I spend so much time so that make the program so easy to use, so that a user can just like download the program, get it run in a second, in a minute. You do not have a lot of, you know, software know that you got it, you know, it's a dependency issue, it's a big productivity issue. I spend so much time so that a user can make use of BSHR for their project. I spend so much time so that I can develop software so that I do not need to support the boring, you know, the boring just like, you know, how to get BSHR set up, go to get, you know, this configuration. Yeah, yeah, it makes sense. Also, just like, you know, after this, when the DSHR have nearly never hear a user question, how to set up DSHR, how to get it up running, that problem by design is gone. To me, it's a relief. To a user, it's a relief. It's a win-win. It just needs me to spend so much effort to get it engineered in a way that it's solid, it's useful. It is robust. It is certainly a deliberate decision. When I did the data based on my period, it was bought in 3DMA. Okay, so I think if there are no further questions, we will thank you again. And for all the participants, I will just mention that we will start again our series in January with the BioXL webinar will start again in January. Please follow the mailing list or the newsletter to know more. It will be definitely a webinar on the GROMACS, new GROMACS release, and there will be also a webinar from ABI on how to use social media in research. And then I thank you everybody. And I take the occasion to wish you a good start of the new year. Bye.