 Thank you everyone for being here and thank you especially to our organizers for making this conference possible. My name is Lauren Harman and I'm excited to be presenting chromophobe on behalf of the Tim Trish lab at the Van Andel Institute. So first off as a little bit of background, we know that eukaryotic cells have two meters of DNA which needs to fit into a tiny nucleus. The way that this is done is that the DNA is wound around histone octomers also referred to nucleosomes much like string could be wound around beads and this is referred to as chromatin. So these histones have various chemical modifications to them such as methylation or acetylation and these are written and erased by different proteins and importantly they are also read by proteins which are able to recognize combinations of these chemical histone modifications and execute a set of instructions based on these marks. So in 2001, Genoine and Alice proposed that these histone modifications could be thought of as a histone code which is conserved from many organisms such as yeast to humans and it dictates or it provides the instructions for which chromatin is wound and unround which controls regulation and repair. Now these histone readers are able to recognize and execute based on these histone marks but for us researchers it can be a little bit more nebulous to interpret this and so because of this Ernst and Kellis in 2010 developed a hidden Markov model or Chrome HMM which can be used to learn and interpret the chromatin states that the nucleosomes are in based on histone marks. And just to provide a little spark notes version of the way that this happens, so the chromatin marks are binarized so they're defined as either present or absent in a given location on the genome and then the user is able to define a finite number of hidden states and then based on this the algorithm determines the emission probabilities. So this is the probability that any given mark is found in a given state and then it also determines transition probabilities. So this is the probability that a state will be transitioning from one state to another. And so the main idea here is that Chrome HMM is able to learn de novo from the data based on the histone marks where we might be seeing an enhancer, a promoter and what might be active or transcribed. And since Chrome HMM has been developed there have been many modifications that have been made to it in order to expand the biological capabilities. One example is a stacked or per cell type model. There is also a chrome gene which is able to predict based on the gene whether it's active or repressed. And what I'd really like to highlight here is SC Chrome HMM which has been developed by Zeng and Srivastava. And so I really encourage you guys to listen to Avi's recording later on. But what this is able to do is it looks at the posterior state probabilities. So what Chrome HMM does is it predicts the most likely state that a genome is in at any given location. But of course with any of these predictions there is a probability that is associated with this. So for some regions in the genome we are more confident than others. And so SC Chrome HMM takes advantage of this additional quantitative information. And so this brings me to our work on our package chromophobe. And of course I am somewhat preaching to the choir here with this audience but we know that BIOC has many power tools that really can facilitate genomic analysis. So using things like our track layer, our SAM tools and BIOC parallel we are able to facilitate the important extraction of models. We are able to use genomic ranges which can make queries easier. So in this example here we are extracting the poised regions and then subsetting based on that. We are also able to use motif analysis packages such as Mona Lisa and Motif Breaker enable to facilitate interpretation of these results such as which transcription factors are being regulated here. And then also null ranges which is going to be presented later on which is really useful for hypothesis testing. Chromophobe is also useful for adding new diagnostic plots and providing tools for working with these Chrome HMM models. So it's able to use genomic segmentation to add a track line for export and course to and from G ranges. It's able to do simplification, recoloring and compression of the HMMs. And then one of the things that we are very excited about is that it's able to plot these posterior probabilities. So in this plot that I'm showing here you can see the posterior probabilities. And you can see that in different regions in the genome we have different amounts of confidence that genome is in a particular state. And so as you can imagine as you are doing your biological interpretation of your data it's important to know what is the probability that a state assignment is actually correct. And just as an example as to how we can use chromophobe we've been working on a project with our collaborators from the Crosig lab and the goal of this project is to understand how dendritic cells which are specialized immune cells are able to undergo chromatin changes in response to a stimuli. So in this project we looked at three different strains of mice and they were treated with either LC-MV which is a mouse virus or PBS which is a control. And then we looked at the RNA in a taxic data as well as different chromatin marks or histone marks sorry. And then using this data we applied the 18 state road map chrom HMM and we really focused in on these bivalent regions and so as you can see from the donut plots as we moved from PBS to LC-MV so the virus stimulated mice these bivalent regions were decreased and if we further looked into the transcription factors using Mona Lisa we were able to see that it was mostly Etsy factors and the NF Kappa B slash rel factors that were enriched and that were really causing this change that we saw as the dendritic cells responded to the mouse virus. And so just to summarize we are undergoing active development for chromophobe. We want to polish it up and get it ready to be published on Bioconductor but as of right now we do think that it is very useful. This plot here shows how using chromophobe it's able to facilitate the analysis and visualization of these chrome HMM models and we're able to see so for example here we can see that triples 2 which is important for dendritic cells is different based on mouse strain and cell type and so this is able to help us to make novel biological discoveries and chromophobe is able to facilitate these sorts of analyses. And I would just like to thank of course everyone in my lab who helped with this project especially Tim Trish and Ava Jensen who are here please say hi to them. And of course our collaborators and our funders. Thank you very much and I'm happy to take any questions. It's either online or in person. If not please feel free to ask questions in the chat and we'll prepare our slide deck for our next speaker who is Jacob Morrison from the Shen Lab. Give me one moment Jacob and I'll put your slides up. Thank you Lauren and please ask any questions in the Webex.