 We'll go ahead and get started. I have to say so this workshop was had a cap of 60 people and I think we actually have far fewer than that right now. The way it's going to work was it was to do two breakout rooms. I think if we have only I mean fewer than 30 people at the end of the day. We're just going to do one session and the way that I wanted to run it was to have either Muhammad or I as the leaders of the breakout rooms take you through the response tutorial. However, if there's just if there's just one big group that I think probably I'll just leave the one big group. In any case, what we're going to do is is go through the tutorial and I'm sort of unlike the webinar style where we have the question and answer and there's maybe hundreds of participants as we only have a few participants. When you have questions you can feel free to interrupt. I don't think we're going to go over time at all. So please just feel free to interrupt at any time with your questions, unmute yourself and just ask away and I'll be happy to answer the question. Muhammad will be happy to answer the question, depending on where we're at. But I was just going to start out with this video this sorry to speak not video but this presentation about exploring the code high C data. So my name is Niva Duran. I am an assistant professor in the lab in the Department of molecular and human genetics at Baylor College of Medicine. I have spent most of the last decade actually working with high C data developing pipelines and developing visualization software and juice box is our big visualization software package is kind of an ecosystem as I'll introduce you to in this little talk. And at the end of this you tutorial you should be able to be pretty comfortable doing things and juice box and we'll hope that you will get out there and start exploring some of the data that's out there, especially exploring the encode data. That's available on juice box and you can look at other kinds of encode data next I see maps. So I'm going to tell you all about how to do that. So, first I hope everybody's familiar with the high C experiment. If you're not familiar with the high C experiment, we do have a background tab with the tutorial that I'll show you that you should go ahead and read that and watch that video. The high C experiment is a great talk about it really concisely explaining what it is and the experiment the discoveries we've made using it over the past few years. But the high C contact maps that the high C experiment generates these genome like contact maps. And so what does that mean. If you think of just a single chromosome this piece of DNA is a chromosome this piece of DNA is represented in the chromosome if you lay out the chromosomes in some canonical order. That is one axis is the whole genome is one axis of your contact map and the other axis is again the linear genome so it's the genome I guess the genome making up the X and Y axes of these genome wide contact maps. And already as you can see here you can see structure from the data so there's things popping out in this case with popping out are the chromosomes themselves the fact that they interact with each other. More than they interact across chromosome boundaries with the other chromosomes and this is a known biological phenomenon, but we can see it directly in the high C map so it's a nice positive control. Now we store these contact maps in multi resolution high C files. Now high C files are the center of the juice box ecosystem. We have software for going from fast cues to high C files which is called juicer you can also go from high C contacts. We have a juicer data archive of high C files is also a bunch nowadays on both encode and on geo. And then we have various tools for getting data out of the high C files, including being able to do feature an annotation and analysis juice box which we're going to talk about a lot today for visualization and then we also have straw which is an API for directly accessing the data programmatically so that you can, for example, do your own analyses in Python or whatever language you would like. So high C files are multi resolution maps, they're stored at varying different resolutions, you can actually be user specified. We usually do like 2.5 megabases down to 5000 base pairs or 1000 base pairs. It's a highly compressed format. It's indexed, which allows for fast query and rack rapid random access. So when useful metaphor for juice box, which is our way of visualized visualizing the high C data these contact maps is Google Earth. When you think about the genome, different features are visible at different resolutions and that's the same thing as if you look at the planet if you look at Google maps and you just start zooming in clicking into zoom in you're going to see different features arising at different scales. So the genome wide scale, you might see for example, the continents and that's equivalent to the chromosomes. Now what might a chromosome look like, what kind of things might you see. So this is just to try and give you more of an instinct of what you're seeing translating topologically into the contact map. So for example, if you have a chromosome and it's laid out linearly ABCDF, if you just did like a DNA experiment for example, what you would see is a really strong diagonal. What that's showing you is that things that are close together in one dimension are close together and three dimensions. This is kind of a normal fact. And so for example, C is close to both B and D, but you don't see sort of off diagonal interactions. Now, supposing that your chromosome was not linear but was a circular chromosome, you would see a different kind of chromosome map. It would look like this you would have contacts also between A and F and so that would look like this off diagonal interaction due to the circularization of the chromosome. If we continue to zoom in to the chromosome view that's kind of like zooming into the country. View, what you start to see are these checkered patterns. So these checkered patterns correspond to what we found to be sub compartments. And they're just areas where the genome is wrapped around itself as you see in these different little areas. And they're interacting preferentially with each other, despite the fact that they are rather far apart on the linear genome. So you can see some of these off diagonal interactions, these areas of red squares. It's just this repeating pattern that you're seeing. And we posit that this is these sub compartments that you're seeing when you look at that checkered pattern. Now, if you keep zooming in closer and closer onto the city of Houston, you start to see squares and peaks. These are strong areas of enrichment along the diagonal they're pretty striking when you do the high C experiment that you see these these squares along the diagonal and these peaks off the diagonal. And what the squares are what are called domains are topologically associated domains, and then you also have what are called loop domains which are these domains that are anchored by a loop in the corner and we posit that they, the cartoon version of this looks like this where you have something that's anchoring them together forming that loop and then causing that DNA to preferentially interact with itself. So then if we zoom in one more time, we're going to get down to the building level to our building at the Baylor College of Medicine. And you can really look right next to the loops and see what they look like. And so this is just to give you this little cartoon example is to show you why we say that this contact map peak is equivalent to a cometson loop. So if you take one side of the genome, it's like this red dot orange dot so you're looking at an off diagonal the equivalent of like an off diagonal box around this peak right here. You have red, orange down to yellow then you have some empty area in between so that would be equivalent to this sort of distance off diagonal. And then you have for B, it's called gray until it goes from purple to pink. And if you looked at what this kind of cartoon loop would look like in a heat map, this is what you would see. You would see that this A and B locus is preferentially interacting with itself that these are also closer than you would expect, given their distance off the diagonal. And so we can also annotate these maps. And that was equivalent to like what you see in this was with Google Maps giving you annotations of the streets. That's kind of like looking at satellite view with annotations. And what you'd see, for example, our encode tracks so we have RNA seek, CT CF, you can look at different annotations domain annotations loop annotations, and you can also draw your own annotations and draw your own lines to get a better idea of everything is located. So there are two main flavors of juice box, there's juice box desktop and juice box on the web. So for this tutorial we're going to go through use box on the web. I do, I don't think we're going to have time to do juice box desktop and we're not planning to do it. But if you would like to learn more I do encourage you to do that part of the tutorial. Juice box desktop has just a couple of things that are different from juice box on the web, which is that it. Mainly is that it allows it has a versus mode which is something is a mode that doesn't exist yet in juice box on the web. You can see a little bit more in terms of data set metrics and sort of quality control statistics. We have this real time visualization of discontinuous regions and an assembly tools module, both of those are sort of heavy duty kinds of interactions that you don't. We don't currently have working for the client side version the lightweight version which is juice box on the web. Juice box on the web does have the very core functionality of juice box, which is the ability to view the contact maps alongside encode tracks like big weight big bed gff etc or really anyone dimensional track in those formats, the ability to load maps and tracks from remote servers customer URLs local files dropbox Google Drive, basically anywhere on the web that you want including geo and encode. It's also has this image that it's this cloud based web app that's pure client side there's no server component. It's embeddable. You can host it on your own site and you can even browse maps on your cell phone. And the thing that is actually most I would say most special about juice box on the web is that it makes it really easy to share exactly what you're doing just by sharing a URL or even a QR code so you can have a poster where you've looked at all this stuff and then put the QR code on your poster and somebody can bring up on their cell phone exactly the map that you're doing exactly discovery you've made. We also have straw, which is the API that I talked about earlier is for rapidly streaming the data without downloading files. You can stream from any kind of server. We have Python JavaScript C plus plus are not lab versions of straw. And we have to burn notebooks showing people how to do some really simple analyses on high C files using straw. These also power the straw, the JavaScript version of straw powers some web based visualization engines for I see data including the wash you genome browser IDB J browse and of course juice box on the web, which we will talk about in just a minute. So I'm going to start the tutorial. I'm going to you're going to go to this website. You'll also have the ability to link from it but this is the website you should go to that's where that's what we're going to start doing and my plan is for everybody to do exactly what I'm doing at the same time I'm doing it and to stop me when they've got questions. I'm in the show right now. And we have 23 participants. So, oh, and I see maybe we have some chat from. Yes, so Mohammed is sending everybody the link which is excellent. All right, so does everybody have a link is everybody have this page up in front of them and they need to get started. You can check it out if you do not. Alright, I will take your silence as enthusiastic participation. So, you were starting with this juice box on the web. We're starting with this lightweight version of juice box. And what you want to do the first thing you're going to want to do is go to HTTPS item lab.org slash juice box you can click on that. What it'll bring up is this blank panel with load map load be mapped load tracks session and share at the top in this plus sign. I'm going to tell you all about all that different thing all that different stuff. So first of all, I clicked on juice box web but the very first thing you would see was the welcome so that's fine. And then the background time I just want to quickly point this out. This is the talk I was mentioning that eras gave in 2017. And it's a very concise explanation of high C experiment and some of the things that you can discover using it. All right, so here we are juice box on the web. The first thing we're going to do is load a map. So we're going to go to load map. In code. And now this is populated with all of the encode maps that they have which as of right now is looks like there's 69 that are available on encode we're actually going to choose the very first one that pops up. Oh, sorry, and these are all the I should clarify these are all the encode maps associated with North of these are all the encode maps. Yes. So, you click on the first one which is MBA one primary plus replicate jam 12878 experiment. It has nine bio reps. So very big experiment has six billion reads so we're going to click. Okay. And was then after it loads up, we're going to want to zoom in on chromosome 17. You can click 17 in the bar right here and that will zoom in on chromosome 17. You can scroll down and go to the box and click there. You can type 17 here, where it says all, or you can also go to this sort of little hamburger menu. You can see there's a chromosome picker right here. So I'll just do that to start. Click refresh next, but everybody should zoom in on some some 17. And once you do that. The next thing we're going to want to do is change the normalization. So we change the normalization from none to balanced. This is a way to mitigate the fact that there are different amounts based on accessibility or DC content and things like that. So basically what a balanced. These are all different normalization techniques. Coverage is one that that simply divides by tries to even things out by dividing by the sum of the row balanced is an iterative algorithm where it makes sure that each of the rows and columns some to the same number. So it basically is a way to account for coverage biases. Now we're going to keep zooming in. We want to zoom into a specific area. So there's several different ways to do that. You can draw, you can take like the option key or the odd key on your computer and press it and then click like that. If you're on a mobile device, you can do the sort of a pinch and zoom kind of thing to zoom in. The way that we're trying to zoom into is in this tab. So, as we scroll down, we are interested in 69.4 to 72.3 mega bases. We can also type the exact location here. So I'm going to cut and paste it to do that but whatever you do you can just go to 70, 70 69.4 to 72.3. The most important so right here you can see it says 69.5 over here 72.0 essentially over here. And then you want it to be at five kilobase resolution. And what you see or there's some interesting seems to be some kind of like interesting things happening here in the genome. So, now we're going to we have finished the first tab. I'm going to just look for a second at the chat. If anybody has any questions. So if somebody had a question, please just interrupt me with questions. So somebody asked what the difference was between balance and genome my balance and Mohammed answered. I'll answer as well. As he says balance just uses intracromosomal reads to do the balancing, whereas genome wide balance is using the genome wide data. I'm pretty sure on this map we don't have it to pre low resolution because it wasn't possible at the time to do genome wide balanced for like single single for kill based resolution for example. Just interrupt me and ask questions. This is going to go way too fast otherwise. All right. So we finished this juice box web tab. You're at this region. We're going to click on loading annotations that's our next step. Now what we want to we're going to do is load some in code tracks. So we can go to load tracks in code. Now this takes a while to load it's going to take longer than did for high C maps and that's because there are more tracks there's no more than there's not just like 70, like there are high C experiments there's you know 10s of thousands of experiments. So this is loading all of the code tracks that exist for this genome which is 19. And what we're interested in the CTCF in our cell line which is GM 12878. And now we still have 50 entries and so we're going to do Bernstein map just because we like them. And in this case, I'm going to load the single the signal p value track of the two combined replicates. We're going to load whatever you prefer to look at when you look at data like this we tend to we like to look at signal tracks in our lab but different people have different opinions about it will look like look at. So, we're going to click that track and click okay. And the track is going to load. And then we can go to load tracks again in code. So we have CTCF, we're going to do H3K36 and E3 and we kept it Bernstein. So we're going to do the signal p value track again. So we can do things to make this look prettier for ourselves and make it easier, we can click this little wheel here to change different things about it. And the track color, I'm going to change it to green, but you can change it to whatever you want. You can change it to magenta or bright purple or whatever you want. I'm also going to change that track color of the H3K36 and E3 and change that to orange. And then I'm going to go ahead and change the name because it's a little long and I know that I'm dealing with GM 12878 so I'm just going to get rid of that and get rid of the other one. But I want to note that, you know, if you're looking at something and these are getting in the way the labels if you just click right on the tracks, they just go away. So then the label goes away and a little thing goes away. So if you were trying to like see at the corners and you couldn't really get a good view you could always click that and you would be able to make it go away. So already we're starting to see some stuff that maybe, you know, something's happening here with the CTCF. There's like a peak here and a peak in the H3K3673 and like, oh, maybe there's a feature here. So let's go ahead and explore this a little bit further. The way to do that is to look at 2D annotations. So they also have 2D annotations from ENCODE. If we're going to load tracks ENCODE again. Now what we're going to look at, we're going to look at GM 12878 in the search bar and then domains. And we see topologically associated domains for GM 12878. So we click OK. And then we can also load some more. We can go to load tracks and code. Neva, we have a comment in the chat. I think it's going a little too fast. I will slow down. Could you repeat the track that you just added? Sure. It was GM 12878 domains. And it's up on the screen. What you see when you look at it is there's GM 12878 originated from GM 12878. This says it's a high C1 topologically associated domains. And I'm doing the combined biorep, which is one through nine. All right. Does everybody, everybody loaded the domains? Actually, it doesn't work for me. I could load them high C data, but it doesn't allow me to put the tracks from the ENCODE. Neither the domains, the keystone mark. I cannot click on the track. When you click on, so you click on the row and it doesn't do anything? No. It doesn't turn to a different color? No, no, no. Actually, it happened with the high C data, but not with the tracks from the CTCF and the keystone mark. I was wondering, I'm working for my Mac. I'm not sure if there is anything. I'm not. It doesn't work for me either. I'm trying now with Chrome because I thought it would be something of Safari. So I'm trying now with Chrome. It's not working for, are you on Chrome or on Safari? I'm on Safari now. I'll try from Chrome, I think. Thanks. Sorry. You guys are debugging for us. I had no idea we had a Safari problem. That is not good. Yeah, I always use Chrome. Yeah, it works now. Okay. Thank you. This was, this is an important bug for us to fix. I'll give you guys a second to catch up. I can talk a little bit about our algorithms. So these topologically associated domains are called, the ones that you see here are called by our algorithm called Arrowhead. There's other different algorithms that people use to call topologically associated domains. So I'll just note that, I mean, we're doing this load tracks and from encode, but there's other things that you can do. Like, for example, you can load from a local file. We have some set 2D annotations that we just have in a menu here under 2D annotations. You can load from a Dropbox file, a Google Drive file, or URL. So you can just make your own, again, you could do your local file and it would go into your computer and load it. But if you wanted to share with other people, you probably want to put that on the URL so that you could click share and they could be able to see what you're looking at. But all that is to say, if you know you could also compare different people's annotations on the style line if you wanted to, if you weren't that happy with the Arrowhead that I've shown you here, you could always look at other people's annotations as long as they were in the bed P format, which is the standard format for these, you could always load them up and they would load onto the contact map. So I'm going to do load tracks and code again. I'll do GM12878 long range. I spelled range wrong. So GM12878 long range chromatin interactions. And again, I'm going to do the combined replicates. So it's one through nine on the bottom here. You click OK. And you should get little cyan dots appearing. And so these are loops that are annotated by our algorithm that's called hiccups, which is a loop finding algorithm that looks for areas of focal enrichment compared to the surrounding area. And I just want to double check one thing. Okay. So I'm going to go so I'm going to go ahead and put this into the chat. So where we are here, I'm just going to hit the share. And copy this URL. I'm going to put it in the chat. So for those of you who needed to be load onto Chrome, you can go ahead and and click this tiny URL. And it'll bring you to the view that I'm showing you right now. All right. So the one thing I wanted to also point out on this on here is that you can also use the shift button. And if you click shift, where are you finding more bugs? Oh, here we go. If you click shift and then you just pan around in the contact map, you get this cross hairs. So that shows you what you're looking at. And in particular, this is useful for like looking at the the loops like looking at the sign in here and you see the the peaks in the CTCF. You can see, if you look on the X axis, the peak and then what the Y axis or you see the peak there and you can look and see that this seems to be pretty consistent. And you also see that the histone marks appear to be lining up with the domains. So this were discoveries that we made in in 2014 and did some more confirmation on in 2015, but those papers are at the bottom of this loading annotations tab. And, you know, you feel free to to look at those papers if you want to know more details about that. But sorry, I have a question. Can you save the session as you will do in the UCSC browser, you know, to save that the track that you want to visualize associated to this kind of IC or these kind of things. Say it again if can you take this and translate it directly to the associate some high C map with specific tracks so you have like a session that is saved and then you know you open the session and you know that it's already associated some high C map with other with one D tracks. So, we don't have the ability to automatically associate like these tracks with this map is that what you're asking. We have this we have this ability to whatever you're looking at whatever view you're in to do the sharing. So we don't have those relationships from, and I guess it's something that we can think about adding as a feature. Having those relationships known from in code so that you would say like, oh well I know that I already loaded that biosample type. So we actually to have some kind of like suggested tracks that are based on the biosample type, because we do know, like when you load map in code, you have this biosample and this is a very specific object in in code. So that this word says GM 12878 originated from GM 12878 that is the same as these tracks that we're looking at it's it's these are operating on that same biosample and so that is that's that would be a useful feature. We don't have it right now. UCSC does load high C files. So you can look at them in the in the genome browser but they use a different view than we do, which is fine but they use the sort of the triangle view. So, it can be a little harder to see things that that are far off diagonal. Does that answer your question. Yeah. So what is the station. What is the station. Oh, where's the session. Do you want to talk about that. Oh, sorry, let me pull up the stuff. So the idea from session from a correctly is basically that we want to be able to, you can save this sort of session, if I recall. This is a newer feature actually that I haven't utilized as much. But let me, sorry. That was my question. Yeah, let me let me get back to you in a second. Sorry. Also, there was also some questions on the chat as well. One second, I'll be right back. Okay, I will answer the other question on the chat, but please feel free to unmute yourself and just interrupt me. I guess I should stop more frequently and ask a few questions. So somebody asked about compartments. So the way that we classify compartments usually is using an eigenvector and so you can usually load. I can look and see if it's in the frequently used here. Yeah, no one wants to get us. Yeah, it's not. So in juice box desktop, you can load the eigenvector track. It'll be calculated directly from the data. We don't have that capability in juice box on the web because you're not, you need the whole, you need to load all the data, like the whole, all of the data in order to calculate the eigenvector. So doing it on the fly is not really possible, except at low resolution and that's certainly not possible with the client side Java. Sorry, the client side JavaScript, but you can calculate it offline and then load again, as I said, you know, with the URL. So you would just load the eigenvector track and look at positive and negative values to do compartmentalization. The other thing to do is to associate that with, like, the methylation track to know, I mean they're very closely associated with things like the methylation track to know they and be part. Sorry, a quick thing about the session aspect so basically if you save a session you can also save it instead of saving it as a URL that you can share, you can save it as a JSON file. And this was primarily intended for sometimes you know the URL you have so many tracks loaded so many customizations of colors and to the annotations. Maybe and several maps like Niva will show how you can have multiple maps. Sometimes the URL gets so big that sharing by a URL may give some problems and so saving by a JSON file is just another way of doing that safely and being certain that everything was saved exactly as you wanted it. Okay, so it'll save it as a JSON file and then you can do this like you can share that JSON file with your colleagues. You can share that JSON file you know on Dropbox and so on. So it's another way of sharing outside of the share URL link. Thanks. So for the next thing I wanted to talk about was comparing maps and the first thing we're going to do is go to chromosome 21. So I'm just going to select everything in here and click 21 to go to chromosome 21. And we're going to zoom in. I'm just going to double click and double click some more. And I'm trying to get towards not 27. So I'm going to pan up around here maybe double click. And actually, the area that I really want to look at is a gene so I can put the gene in so I'm going to delete all of that and type. ADMTS1. Click return. And it goes to that location. I'm just going to scroll over a little bit just because my screen is a little bit small. So now, sorry, could you repeat what gene you're looking at? So it's ADMTS1. Thank you. No problem. And the other way to zoom into where I'm talking about anyway is it's about 27 million to 30 million on chromosome 21. So one of the most useful features I think of Juicebox on the web as opposed to Juicebox desktop is this ability to prepare maps and look at them side by side. So if you click the plus button, what you'll see is you get a map panel. Now, this new map panel has a black line around it. And that's because that is the one that's currently live. And if you load a map or tracks or anything else, it'll go into that panel. If the black line is instead around this MBO one primary replicate GMI 12878 on the left, then anytime if you load a map, it'll load here. So you don't want to do that. So don't do that. Make sure that you have your black line around the empty map panel. You have map in code. And now instead of looking at GM 12878, we're going to look at I am our 90. So it's I am our and then a hyphen, and then 90. And again, what we're interested in is the combined replicates. So you're going to need to scroll down to see the combined replicates. But they're down here with the bio rep one, two and the tech reps. If you click that, you click okay. And you sort of already immediately see that these are the same regions, they look pretty different. The map is shifted up a little bit. So we're going to fix that in just a second. First of all, when I look, the first thing I always do is do normalization. So go ahead and make the normalization balanced. And then why don't we go ahead and load tracks into this map the same tracks that we've already loaded. So I'm going to go to load tracks in code. Now instead of GM 12878 I don't want any of that I want I am our dash 90. And I click okay. And I see. So somebody saying that the plus is not giving them a new window. I'm not sure which browser. Yeah, so does share plus sign supposed to give a new parallel like the not the share button, but the plus button plus button. Yeah, but then it doesn't, it doesn't generate another not maybe just me but it doesn't give me a new side by side view. It's below the map because sometimes the window is like too big and might put it below. I saw something below if I scroll down. Okay, thank you. So now if you're going to load tracks and code sorry and I typed the I am our 90 and then I didn't do anything with it. We're going to do CTCF. And we have several possibilities here. I have just like marked the ones that I decided to use in this. So it ends up being on the second page. This signal p value track again from Michael Snyder lab, the ascension number ends in IZF. You can always search for the ascension number if you want. That's also in the tutorial you can copy and paste it. But if you click the signal p value track, you click OK. It'll load the CTCF. I'm going to go ahead and do the same thing again load tracks and code. And then we're going to load h3 k36 me three. And again we're going to do signal p value and it's this one that ends in a AC this combined replicates signal p value tracks we click OK. And then I'm going to go ahead and and change my track colors again so this is going to be green again. I'm going to change the track color of h3 k36 me three to orange. And then I'm going to change the track names to CTCF and h3 k36. Now if I want to get sort of apples to apples comparison I also want to load these these two d annotations. I'm going to load tracks and code. And so for I am R90, I'm going to go I am R90 domains. Again I'm going to click that. And I click OK. And we'll see the domains load and then load tracks and code and I keep doing this like load tracks and code part but you can actually select multiple at the same time. Here we have again I'm R90 long range chromatin interactions again these are what what we call peaks in the contact map. I bet that's what in code calls them as long range chromatin interactions. You click that and you click OK and those load as well. So the next thing we're also going to look at is load tracks. Now we're going to go somewhere a little different, slightly different. We're going to go to frequently used and we're going to look we're going to load the genes, the gene track. I already did that for the Institute MBO one I am R90 experiment so let's click over to the other one. And to the MBO one again. And I see we have more chats. I just want to make sure that. Yes, I will sit down. Here you will also load the genes. So again you go load tracks frequently used. And then genes. I just want to be clear that all of this is in the tab. Everything I'm talking through is in this tab in the aid in lab getbook.io. Basically everything I've done is all here. So, obviously I would love for you guys to just be doing this at the exact same time as me and I, and I'm happy to slow down to make sure that you get there. But also if you're feeling very lost again we have a tiny URL that that's here. So this you should see something like this tiny URL dot com in the comparing web, comparing maps web on the tutorial if you click that you will get exactly what I'm showing you. And you can just start from there instead of, you know, if you're feeling very lost. But I will give a minute to catch up. And I will look at the chat and please let me know if you have. If you have any questions I'm happy to chat with you go right now. One more thing I'll mention on the website that you can see where I'm using the juice box you can see that there's this link called forum. The forum link is a Google groups link. Mohamed and I often answer questions on those they go right to our email. And so there's also a lot of questions from the past six to years or so. So please feel free to ask questions there if you are unsure and want to know more also, of course, ask in the slack for the encode 2020. We're happy to answer questions there as well. You ask it. And then we're exactly as the load tracks genome annotations, if you could. They go to load tracks, and it's under frequently used, I know it's a little bit confusing. So instead of encode it's under frequently used to click frequently used. And then if you just click down, just click in that bar, it shows jeans is the first one that shows up. These are a little bit funny because they're. I mean these are all also on encode pretty much, except for CTCF orientation. And it happened to be the one the specific tracks that we used in our 2014 paper, which is why we had them under frequently used but and also jeans is not on encode. I have a question. I was wondering because how big are these high C files after the research analysis, because there's many, many resolutions, I guess that I don't work with him. So I would like to know how big they are. Is it possible to work on locally on your computer once you analyze your high C. You know what I mean. Yeah, it depends on how much space you have on your computer. I mean, I mean, sequencing that let's say that you don't know what big one year one million when you read so this is so this is 6 billion reads. Yeah, this map is 6 billion reads this map is 2 billion reads I think something like that. They. So it depends on what you include but if you don't these days we so we used to include what are called fragment limited resolutions we've moved away from that if you include fragment limited resolutions on your maps so you're just including like one kilobase up to 2.5 megabases, then they're around like, I don't know 12 gigabytes for the 6 billion read map for if you have only a billion it's smaller than that. So, I mean, again, the reason that we have worked so hard to have this structure where you can browse these very easily even over the web. Without having to download the whole file is because the files are big but they do have this nice index structure that makes it pretty easy to access and all of our tools. And it allows you to stream so straw as well anytime you have a data API you can always just have it on some server that has more space than your, than your hard drive. I do have like some high C files on my hard drive, but I have like a 500 gig hard drive on my Macintosh so I mean it is a signable percentage but it's not. You know it's a few movies worth for sure. So from the server, the data on the server. These are all on the server. We're browsing. Okay. Okay, thank you. Yeah, no problem. I might have to go and, and plug my computer in. Mohammed warned me that I would have to do this but I didn't think I, I didn't think I would. But here we are. So, I'm just going to finish this tab and then I think I will go plug my computer in and then come right back. So, we can also bring in additional encode data. So we're going to load in the RNA, the RNA seek signal tracks for H3K4 ME3 for both GM12878 and I am 190. So again, I'll go, I'll try to go up slowly. I have a very bad reputation for going too fast in general on everything. So when you're an MV01 primary replicate GM12878 make sure that that's where your black highlight is. Like load tracks and code. And here we're going to load GM12878 GM12878 and then we're going to do RNA seek and there's still a ton of entries. So I think that's, I think that's why it was a big reason that I put that the actual. So I guess I want signals. We can do RNA seek and we can type signal and that gets us to only 10. And so then I think the one that I was most interested in was something that ended with SCA and the ascension. Oh no. It's still 446. So I'm going to just take this. I'm going to type the thing, but you should copy it from the tutorial. Enc ff 001 SCA that gets us this track. My Mac is really warning you now. So I'm just going to leave this up for a second. So people can see that. I have a quick answer to the question. I had that basically. Yeah, there are more loops and domains in that region you're at MTS one but also if we look at the boundaries of where those some of those domains fall as well as where those loop fall loops fall. They line up with the Adam TS1 gene as well. So, so it's not just that in that general area but also at that exact boundary. We see much more activity of those loops and domains. So I've loaded my GM my GM 12878 or any seek track. I'm going to click over time or 90 load tracks and code. Now I'm going to do. I am our dash nine zero. It doesn't like my lower case. I think maybe it doesn't like signal. No, it's because of the ascension at the end. So I get rid of the ascension, and then I still have like 164 entries. I'm going to go down and look for the actual RNA seek signal, which is this Inc. F F 0000 HAN. So I encourage you to do the same. Go to your tab that has the comparing maps web. Scroll down to this area of let's explore these differences by bringing in additional encode data and then copy the I am our 90 ascension. I will perhaps show you two things here so I'm going to do that ascension and I'm going to click it. And then actually, after I've clicked it, what I've done is just kind of like a checkbox so I can also highlight in the original get book. The next ascension number, which is ENC F F 254 FBR highlight that copy it. Go back here, put it in the search box and click it. And then when I hit okay both will be loaded. So what's underlying that. Right there and so but it's going to load it slightly differently on GM 12878 but that was the risk I take by doing things slightly differently. So you can see the I'm our 90 the peaks. Again all on this Adam TS one and if we go ahead and look GM 12878 does not seem to have much let's if we click it and again load tracks so we're clicking over. MBO one primary replicate GM 12878 that's where your black box should be. You click load tracks in code. Now we're going to get rid of the ascension that we just looked at. I'm going to go back. Look at my ascension. So, let's explore these differences RNA seek signal and H3K for me three GM 12878. I've already loaded this SCA one. So I'm going to load this WPV one. So if I click that. Here it is. It's an H3K for me three track from GM 12878 I click okay. And basically there's just, there's just not much in the GM 12878 and there's these peaks around of activation around Adam TS one. So Adam TS one is a gene that's known to be active in I am our 90 and inactive in GM 12878 and as you can see it appears to be causing this difference in chromatin structure, which you can see via these differences in the loop being the loops that are called, but also in your differences in these in these little domains and it's not the case everywhere of course I mean the overall actually the cell lines are are often quite similar, but this is a striking example of where they're different. And this is also an example of how you might look at those differences and be able to scroll around side by side, see the tracks see everything move around at the same time. Any questions. Yes, I have a question. Is it possible to do this kind of analysis between chromosomes different chromosomes. Not in so you can do it. I mean, yes and no, you could go to, if you were interested in a particular inter chromosome interaction so for example if you're looking at chromosome six and chromosome 14, you might have a six here and a 14 here. So that's the reason why this is repeated is because it's showing you like the x and y axes of chromosome 21, but you could do chromosome like six and chromosome 14 or whatever you wanted. But one of the things that we always do in juice box on the web is when you're looking at a particular location you're looking at that location in all of your panels. So you wouldn't be able to, for example, have chromosome six here and then chromosome 14 in the panel to the side. Is that what you're asking? Yeah, actually, I was asking if in the same sample, I can analyze inter chromosomal interactions and then have like in the other side and a different map with from a different sample. But with inter chromosome? Exactly, always comparing like, as you say, chromosome six and 14 for one sample and then six and 14 for the other sample. Just to compare between samples if the inter chromosomal interactions are the same or they're changing. Yes, absolutely you can do that. Okay. I mean, when you load the map, I just want to make sure that we're sort of done with this example before I, well, I can just open a new one I guess. Do you want to share the link for that one? That way we can have it saved in the chat. It's in the, it's in the gift book. True. But yeah, I can put it in the chat. It's probably all pretty color in that link as well. Go to the chat. You can compare somebody asked how many maps can compare at the same time you can compare many, many, many as many as you want. I'm not sure what when things would start really slowing down. You know, we've done six, eight, 12. So there is the loop. Sorry, there is the link. And all right, so let me just randomly like, I'll just load that map again, just for fun. And the first thing that you when you load is you see this chromosome of you. So say you were excited about chromosome six and chromosome 14 you could see six 14 you could double click. That's your inter chromosomal map then if you load a different cell line. If I go load map in code. I don't know. I don't know what I have nothing to heal I could do something else. But just since we're doing that when it loads it's going to load six and 14. So that that's a way that you could look at differences inter chromosomally. There was just one more feature I wanted to show with the with the loading as a control map. So instead of this, which was, which was you can also load a second map so in juice box desktop the primary way that we have, we don't we do have a side by side view but it's not as seamless as this one. The primary way that we have that we like to look at mass and compare them is by comparing them as as a control map or in the language of juice box on the web it's this V map. Instead of using the plus button. You would do load B map so now I'm going to do it and then be a one gym 12878 and I'm going to load the MR 90 again, but I'm going to load it here. And we'll see how badly it breaks. So if you click I am our 90 this one down here. You click OK to load B map. It's going to load it as a B map. Now it says a B and you can, you can be looking at a, and then you can look at B. And it'll change. And you can go back to a and go to be. And this is basically how when we look at things in juice boxes a lot of times what we do we do this or what's called the versus mode, the versus mode shows you a different one on on the other side of the diagonal but we don't have that mode yet for juice box on the web. You can also just click this one to toggle through them. You can go up and down between a and B. And you can also actually make like a little loop. This is what I'm a little bit concerned is going to break but we'll try it. So you can click cycle maps and it'll like play you a little video. So that's kind of nice for like, if you're giving a presentation and you want to show people that there's differences and you're just going to talk over it. And people can just keep looking at it. You don't have to keep toggling for them. It'll just cycle for you. So I'm going to say again how you did the the loading of this B map. Absolutely. So you go to so load map is for the primary map the a map and also when you click load map and you've already got to be loaded I think it's going to kick out to be although I'm not 100% sure. But when you click, instead of clicking load map you click load B map and you have the same options that you would in loading the map. It's basically like, you can load a map from anywhere and you can load your B map or your control map from anywhere just as you would load the regular map. And so this is instead of having the plus and having it to the side, you just click load B map and again I did encode the same kind of thing we've been doing I am our 90. And then I clicked this one. Are there any other questions. One other feature that's kind of neat is that if you if you share a link, like here's an example at the bottom, if you did the share link, I'll just show you just about to show you in a second how to share links. If you click it in this cycling mode, if you put that cycling mode on and you share a link in that cycling mode, it'll cycle as well for the person who's who is loading your maps so they'll actually get to see what you're trying to show them with the cycling. All right, I saw again that there was a something. So, I'm going to go ahead to sharing URLs. This is again, I think I mean, I feel like we started with juice box desktop that was our worker horse for a lot of our discoveries. And we moved to juice box on the web, more recently, possibly because areas a didn't he runs our lab really likes looking at juice box maps on his mobile phone. I mean he really does. But in any event we've really gone. So a lot of the analysis of juice box on the web. This is work that was mainly done by Jim Robinson who's also the author of IGV extraordinarily talented programmer researcher, etc visualization software developer. And one thing that I really like about juice box on the web is that you can do this sharing URLs and it's very easy we did have some we do have some sharing capabilities in juice box desktop but it's not nearly as seamless. So I've already showed this to you before, but you know I can just show you one more time, which is that you can just click this share button anytime and it'll give you this new tiny URL in that tiny URL is encoded the whole state of everything here. Now, the only thing that you have to be careful of is that if you've loaded tracks from so you click local file. This is going to bring up like you're going to get to see my downloads folder. It's going to bring up your, your laptop, and it's going to show you like all these things that are on your, your laptop that you could load as a file but if you try to share that like it's not going to work. I mean it'll work but that file won't show up, because that file is only on my computer and we don't like automatically upload it or anything to a server. So just be sure that when you do a share, and you share this URL and you can share it. So email you can tweet it out to your followers or you can have a QR code. We like to put these QR codes on our posters these days. Then people can just go right to what you're looking at and it's a good way to show them just be sure that you don't have any local files in there and it'll be it'll be seamless. The other thing you can do is you can share things that are like password protected or whatever on your own web server it's just that the other people would also have to have access to that. And then you can embed here is like an example of embedding code that you might use and just showing you the QR code. Now, let's see. Any questions about sharing before I move on. I'm going to quickly talk about hosting with Dropbox driver geo. So you can use we use Amazon S3 actually like most of our our high C files and the juice box archive are on S3. But you can also use Azure or other file hosting services, you can use Dropbox and it's very easy to set up. I'm not actually going to go through this example. It's and I'll read through it with you but I we won't actually do it. Basically, what you can do is upload files to Dropbox and then you can copy the Dropbox link of your file that you've uploaded. And then you would just go to here you would go load map URL and you would just put in the link from Dropbox. Now if you choose Dropbox file it's going to load your Dropbox folder. And then your drive file will load your personal your Google Drive. After you load it, you can just do everything that you would normally do you can do the same thing that you would do again with the loading tracks onto that file and you can share it just as you would. One thing I do want to show you though to go through an example is is loading files from geo. So these days, it's not 100% true but a lot of times when people publish experiments in high C, they will push their process files to geo. So maybe you're interested in seeing some of that. I'm going to load a new tab here at leaninlab. See, leaninlab.org slash juice box just wants to go to the get book every time I'm going to load a new juice box to show this but you can also do it in the one that's already there and just like a little bit attached to to what we have here but you can also, you can do it here if you want instead. What we're going to do is go in your tutorial, you should go to the hosting with Dropbox Drive geo and you should scroll down to loading files from geo. I can also put that link directly into the chat just for fun. Every time I do that. Okay, it's over here. Okay. So if you go to that link that I just posted in the chat, which is, which is the same as this, and then you can look at this series. So I'm going to click this link and open a new tab. What you see is a paper about cohesion loss, eliminating loop domains in high C, and it has a whole bunch of experiments. It has a whole like long series here of high C experiments. These are all, you know, the, the SRA files that are that are the compressed FASCII files. And then there's also process files. So the process files here are high C files. And so for example, I'm just going to choose one. So I'm going to choose this withdrawn one. And so what you need to do is you don't click the HTTP one because of the way that it encodes. You would look at the FTP and you would right click it. So I'm going to do control click and copy link address. Now I go to my juice box window and I can click load map URL and enter that URL. And it will load it. And then now I just have this map. It's streaming directly from Geo. I didn't download anything. And I can just do what I would do with any other file. So I can just click on someone look around and maybe I'm excited about this particular experiment. And different things that we could do with it. You can see the normalizations available. You can load tracks into it like you would load any tracks from anywhere and do all the kinds of analysis you would normally do. But the file is that's being operated on is hosted on Geo and you can you can load it directly from Geo. Are there any questions about that? I think just as a quick clarification for like to about the Dropbox and Geo aspect like this is one of those cases where I think I was mentioning in the previous tab. If you have a local file, you can't share that link because that that file is only on your computer but it's very easy to upload it to Dropbox or Geo and then it's very easy to share with the library. Sorry, I thought there was a difference between Mac and when I tried is it copy link location that will work too? For here, I do on the FTP I right click it and I clicked copy link address. I think that I don't see that on on a PC. I see copy link location. Try that. I should then go to the go back and say load map right? Load map URL. URL and just put this in. It doing something I'll tell you if this is. It's trying I don't know. It's actually worked. That's good news. Thank you. So, can you go back? Like, is that like, you know, undo or something like that? Oh, no, you know, this workshop, people asked me that too. We don't have that capability yet. It's definitely high on the high list of features just having like a one. Like an undo basically. And if you go back with your browser, like it's not going to work. No, no, it's disaster. Good to know. Yeah. I'm not sure that they would be able to do a full. State go back. Although I think it's totally possible because we already, we have all this encoded in the share state. But one thing that we did say that we will try to push out ASAP is essentially go back to location. So sometimes you just click somewhere and you didn't mean to and you like lost what you were looking at. And that can be very frustrating to try and get back there. Yeah. I don't know where I just was a second ago. So yeah, it's definitely on the feature list. I think it would be really useful. Like, one more question now. I mean, when I look at what I did, I lost all the uploaded tracks. Is that normal? If you open. Yes. When you load a new window. Okay. You basically lose everything, right? You have to start from, from the beginning. Okay. Yeah. So you might not want to load the same tracks. When you're, when you're loading a new map. Yeah. I just did it on whatever I had already on the first one on the left. I just wonder, it doesn't matter. It's like, I can reproduce it. I just wonder if it's possible. Yeah. Yeah. It definitely clears out all the, all the tracks you have loaded. But for instance, if you were now to add another one. That copy. At least location. And then you can. Load this track and then you will need to, to, to load the encode tracks again, right? That's what it is. Yeah. So if you add a map and then load a map in, it'll go right to the location that you're looking at. Cause it's the map panels are all tied together with a single location. So I'm just going to show one thing, which is this visualizing from the encode website. So we also have a way to visualize directly from the encode website. And I talked to them about it. I have additional features. I'd like them to add. But one thing we can do is, is actually go to encode project.org. So I just clicked this link. And here it is. And we can go look around for IC experiments. So let's see. I decided that you guys are going to look at each Mac experiment. I think that was just a totally random decision on my part. But if you search, I mean, the other way to do it is you can go to the matrix data experiment search experiment matrix. It'll bring up the matrix. You can look in the assay for high C. Click that. There's a bunch of different ones. You know, you can do whatever you want for this. But what we want is ones that have. High C files associated with them. So if you scroll down, you're. In this experiment matrix, you have assay type assay title. And as you're scrolling down, looking on the, on the left-hand side, you see different kinds of filters that are on homo sapiens mouse, et cetera, et cetera. And you can scroll down to available file types. And, and click high C to make sure that you're looking at experiments that have high C. Files associated with them. And so that limits the assays you can look at. You can decide whichever you want. There's obviously some G1 to eight, seven, eight, because it was the cell line that we used in the 2014 paper and that we've done a bunch of experiments in, but there's also more. Actually, you can see that there's all these different ones that you could look at. So let's just take this. Any one of them. I'm going to choose memory. I've left the cell and I'm going to click the first one. And as I scroll down, you can see that this hopefully you're familiar with this, but if not, you know, I definitely encourage you to sort of like play around on the encode website. It's, it's nice. It's a little, it can take a little getting used to just like anything else, but it's a really nice website. And they have a whole lot of information on here. So here's our, the experiment is telling you how the process files were created. And if you click on file details, it shows you at the top, the raw sequencing data. So the file details is the tab that's over. It's all the way to the right. And you can scroll down and you can see the high C files here. There's, there's just, I mean, these aren't going to be very different because one is just a map and quality threshold or chromatin interactions and one is the mapping quality threshold or chromatin interactions, but these are checked with the visualize button checked. And so then if you go up here to visualize, it says HG 19 juice box visualize at the top under the file details tab, you can click visualize and it's going to open up the juice box and it's going to load those files into it. So these maps are pretty similar because they're basically the same map with a different mapping quality threshold and you can see that there's a different mapping quality. There are differences that you can see and these are, these are due to, these were repeat regions that were jackpot and you can see that there's like a little red thing here for the one that's lower mapping quality threshold. So those, those kinds of artifacts go away when you, when you make sure that your, your mapping quality is high enough. But yeah, so that's something you can do directly from encode. And that's a, that's nice. They've added this feature into encode the visualization feature. What I would like them to do at some point is for you to be able to put it in your shopping bag. So you can always like click these and put them in, you can click things and put them into your shopping cart. They have like a shopping cart thing. But you can't do it on the level of individual, like I don't, like if I put it in my shopping cart and then I go to my cart. I can't visualize directly from here. And I'd really like to, I want to be able to visualize from here. They let you download, but not visualize. So. Can I ask you a question, please? If you don't mind. Sorry. Can I ask you a question, please? So now this, this data that are high seed data that are available through encode, right? They will just give you this interaction map. But, you know, Ted's CTCF loops and stuff. This should be another type of a file though. And are they available? Those are available. Yes. So I'll show you. I had filtered on high seed data availability, but there's also. If we undo that. If you filter based on, for example, bed PE. So bed PE is, is often the loops and domains. One thing that I do need to explain about this actually. So there's, there's far fewer. Which is a good, a good observation. But these would be this, this file that I just clicked on this bed PE is, is the, let's see. We'll go look at the experiment and I can show you more about it. When you look at the association graph, you see that it has these replicates that goes into create these IC files. And then the mapping quality thresholded. High C file creates both topologically associated domains and the chromatin loop identification creates the TADS and the, and the loops. Now. In code. We'll be processing. We'll be doing uniform processing of all the high C experiments. But right now. I see experiments. There's a lot more of just raw data than there is a process data. And there's definitely a lack of process data for the TADS and the chromatin loops. And so that'll be coming online. It's not available yet because the processing uniform processing pipeline isn't done yet. And so they haven't started running it. On all of these assays. But you know, I would say. Over the next few months, you're going to see these, this data be processed by the uniform processing pipeline. And there will be high C file. And then also the TADS and the loops from everything that's in encode for high C. Okay. Great. Thank you. Also, are those data linked? I mean, you need to first load the. Interaction data. And then add TADS and loops rather than just have a loops or it's possible to just have a loops. So. So I feel like I have two answers to your question. So I will try to, to do two answers. So. Similar to the earlier question, I think it's a really smart idea to associate. Data directly from encode with, with the bio samples. And we should do that. We don't right now. So it should be the case that when you load a map, after you've loaded the map, it would be nice if there was some kind of like load tracks. And maybe something that appears that says, you know, here are some suggested tracks that are, are directly associated with this data type from encode. Because as, as you say, we know that, I mean, it's coming out of those files. So we actually know what loops and domains are associated with even this particular example. So that's a great suggestion. We should definitely have that as a, as an addition to the. To the UI here. The second part about the loops, you can just download the loops. So, I mean, in encode, now I'm, I've like lost where I'm at, but say I go down again to this, I'm doing high C experiments. And I'm doing, I want to look at bed PE files. If I go here, I click on this, have one experiment. And I go and see my association graph. I can either do it from the graph or from the file details, but I can click, for example, this is long range combative interactions. I can click that and here's a file. I can download this file. This is the small file. It's 518 kilobases. It's just a bed PE file. And I can then load that file wherever I want and do whatever I want with it. I mean, you can load it into juice box, but you can also load it into anything that reads bed PE, you know, UCSC, IGV, whatever genome browser you want. You can also just look at, like bed PE is just a format that it's a text-based format. So you can also just parse it with whatever you want, like your own code or like Python code, AUK code, whatever you want to look at those interactions. So the data is available. And you can load it separately. In this case, you, you could look for this ascension in our encode browser and it would show up. If you wanted to load it as a 2D interaction. Or you could always download it and look at it elsewhere. Does that answer your question? Yes, it does. Thank you so much. Thank you. All right. So. Does, are there, are there any other questions about this? I'm just going to talk about one last thing, which is this, this interactive figure. So. One thing that we, it was important to us and it's sort of inspired. A lot of the work we're doing is to make sure that the, the research and the observations that we're making are really shareable and verifiable. So it becomes this thing that you can easily verify these findings. And so basically you can, you can recreate a figure of ours. So we have a few minutes left. So what I might do. I'm going to click this. See what happens. Right. So I'm just going to click the link. But this practice interactive figure, I encourage you to do on your own. If you would like after this workshop. But this link, these are these, these figures that we've, these are figures from the paper. And you can make interactive versions of them. So you can, you can click a little, you know, you can do this. So I'm not, I actually don't know what happens with these. I'm, I'm worried that I'm going to click them and they're not going to work. Maybe I'll just see. Oh, maybe they will. Oh, no. What happened with this? So. Basically. One of the things that you can do is you, we have this figure where you see. And what we're showing here is that these. These treated maps where you're treating them to. Remove. This is basically an experiment that's showing what happens. That's trying to show what happens when cohesion is degraded. That, that things look very different in the treated maps versus the untreated maps. And so this exercise is practice interactive. Is showing you how to recreate these figures. And in particular, you can take these figures and. And. You can look at them and see if we're, if we're legit, if we're actually doing something that's reasonable. Or if, you know, we were cheating on the color scale or something like that, like here, I, we say that the red is 42 and this red is 38, but maybe that, maybe we cherry picked it in the area. Maybe it doesn't look like this across the whole map. Maybe this number is not a reasonable number to choose for the red. So, you know, they're not really representative. So this exercise, the practice interactive figure exercise is an exercise to show you how you would make. The figures from this paper. And we actually used this. I think there's a permanent link in the, the paper that this was published in. You can look at that permanent link that's in that paper that was and so you can actually look around and see, play with the color scale, pan around and zoom and make sure that you actually believe the result as it was reported. So Neva, actually if you go back and click that link, it's very easy to change the black to red. It's just a click on. Yes, I think it must be. So this is of course a new version of JavaScript but you can just change it to red and it works then. Sorry, click on red again and just click on red again. Yeah, it should change at least on my window too. Yeah, I think it might be a new bug that we need to fix but the result is still the same. Always fun to find new bugs. Yeah, I think recently there was a new push to add this dark color mode so people can use it at night and things like that. So I think that must have broken this particular. Yeah, and these are old links. I mean we should link. Yeah, these are old links, that's true. We should go somewhere else but that's fine. So that was the untreated and then you can click this for the treated. But the thing that you'd really like to do is have them side by side, which this is red again, which is what the interactive figure is going to do for you. You can sort of see just by doing tabs but this is exactly what we were trying to get around when we developed just that. And these intensity in color is really set by you. There is nothing particular that you know there is no recommendation of what would be the best one or like you know what people would normally report. You just you should always report it though. Is that correct? That's correct. You should always report it. Things do look very different. I mean as you can see when I click. Yeah, you should always report it. There's you know sometimes there's sort of an estimate of like 95 something like the 95th percentile essentially of like you're you can try to figure out or 90th percentile you can try to figure out in your window what your counts are and make sure that like dark red is going to be like 95 percent of things are going to be below that. But these are just the one of the reasons that we have the color sliders because this is these images have very large dynamic range. The diagonal can be very can have really really strong high values in it and so you sort of need to let's see see how high we can go. That's 40, 78, 156. So you even have values on the diagonal that are above 300 here in this map. If you just did this you would look like there was nothing but we know that there's there's things in this map to go back to like 12. You see a lot more. So yeah I think it is it's important to report it and I think it's important to be able to have an interactive way to share what you're looking at so that people can determine if if they really agree with your conclusions. So I did share one example by the way in the chat of a an example of figure 2a where all six of them are cyber sides. This is one of those permalinks that we had in the GitHub version. Since it's an earlier version of this it doesn't have that same color scale changing like in this layout of the three byte or two by three grid. But the newer ones have so if you were to make a newer figure though with the latest version of justbox.js you could have all of those functional and this is also an example of embedding. So this will load in a second but this will have. For tomorrow we're going to change the links and have it be this. Yeah this is great. So these are all embedded and then you can move around in them. Is that the idea right? Yep yeah we can browse the maps yeah. Yeah so yeah once the six maps load you can browse around and move up and down the diagonal and see that it's showing what's expected. And yeah we can update these links in the in the tutorial so you guys can always see the live the latest version. And tutorial will be available forever or on that link? Yep yep yeah so we haven't yeah so we we will leave that as a. So the desktop version so Mohammed you do have an updated don't we have like an update flag that comes out when they. Yeah so the yeah the jucebox version assuming you have the one of the recent ones post I think 2016 if you have it'll actually let you know when a newer when a newer desktop version comes out it'll when you start it up it'll say there's a newer version available would you like to download it for that it will update and it'll let you know to don't. And jucebox on the web it just gets automatically updated on our end. As to how often we update things so we haven't done a big push on jucebox desktop in a while but we have been constantly updating it so I think there's going to be a we're going to have a big release soon with some of these some of this stuff. I would say I don't know what do you think every couple of months Mohammed we we probably yeah usually every couple of every few months I think because so technically speaking all of this is open source it's all on github if you guys want to be using the latest developmental version that you know has everything fresh off like the freshman like the pull commits and things like that then we that's it's possible to build that directly we have instructions on our github for how to build the the latest version that is that's always updated that one may encounter bugs though so that's why we we push every now and then but technically it's all open source so you can always use the latest developmental version if you'd like directly from the the the source code but I think every few months is usually when we when we do it sometimes a little bit more spaced apart before a bigger push okay perfect thank you so much yeah no problem and I think what this was maybe this is also a kind of what you're asking about there's some things that are more heavy duty in terms of their use cases and analysis for example there's some work related to assembly so jucebox assembly tools that specific to the desktop version at the present and there is a it's linked also in this tutorial in this adenlab.githbook.io tutorial at the end it has some information on how to deal with jucebox assembly tools there are a few other tools that are more heavy duty in nature and as such required the desktop version otherwise everything else is basically supported on both the web and the desktop okay thank you so much I was wondering what kinds of applications you would use the the api straw for yeah great question um let's see you can share my screen again hopefully I won't get kicked off um so we have a oh man I shouldn't just do you guys get to always see what my history is um if we go to the straw I think we have we should have this yes the jupiter notebook is available in the repo and so you can actually just look at the um at that to see it's also in the it's also in this uh tutorial at the bottom but it can give you an example I can show you a jupiter notebook of how you would actually use it so for example um if you wanted to to use straw to analyze the high c data you would be able to to do streaming so it's showing you how to install the library and then how to extract data from the matrices how to print out that data um maybe you wanted to have a helper function to extract the data along the diagonal so that would be this you can plot the data so this is plotting the data in that plot lib so you can get um little figures you can change your color scale if you want if you don't like the way that we display things and you want to display it in like a different color scale you would be able to do that um and then this is um analyzing the data this is doing like a Pearson correlation so we did Pearson correlations um in the 2014 paper it's often a way that people um try to compare replicants not perfect but it is one way that you can do so and so this is just showing you how you would do that um in python to get this you know the Pearson correlation coefficient and p value for comparing two different high c files so if you want to compare between replicants for example um and see what their what their Pearson and spearman was this would be code that would show you how to do it and this is just to give you an idea of like I mean people who can do whatever analyses they want um this is gives you an idea of how you might do these analyses and how you might use use straw use python in order to access the data directly okay thank you yeah so it's it's worth noting that there are pieces of analysis inside of juicer tools so doing things like aggregate peak analysis where it'll add up all of the loops that are identified and seeing if those are enriched or um or for a given list like differential loops and things like that so this is more so like more if someone has some type of novel analysis that doesn't already have an existing tool and they want to develop on that further um they can utilize any of those languages to do so but we do also have a number of existing uh post-processing tools that are available through juicer tool yeah so in particular juicer tools has arrowhead for domains and hiccups for loops which are both pretty popular yep and also has a differential hiccups for differential loops as well as um apa if you want to compare like on aggregate if a particular list of loops is enriched or not for different cell types any more questions well thank you so much for tending i hope was helpful it was very helpful thank you so much and we're we're available for questions so offline as well