 So next up, I'd like to introduce Ruiyi Baal. Ruiyi is a research associate professor of medicine at the University of Pittsburgh and co-director of bioinformatics at the UPMC Hillman Cancer Center. And she will be speaking about a package that analyzes digital spatial profiling data. Ruiyi, I don't think we can hear you at all. Okay. Yes. It got disconnected. Can you hear me now? Yes. Okay. Great. I'll put my slides back on. Very much for introduction. Ruiyi, I think we've seen your presenter view as is it possible for you to switch to the slideshow? Switch screens? Yes. I just actually switched. Does it not show now? Like we're seeing the presenter view with your notes in the next slideshow. Yeah. Oh, that is strange. So let me do this. It worked last time. Okay. Let me try again. Okay. How about now? It's the same, but I think we're good. We can get going. That is too bad. Thank you very much for the nice introduction. Today, I am happy to present some of our work in designing software for analyzing and visualizing the spatial omics data. I have no disclosures. When we talk about spatial omics technologies, maybe three major ones that everybody is most familiar with. The first one is single cell RN-seq that is generated by 10X platform. And second one is multi-spec new fluorescence. And third one is the nanostream digital spatial profiling. And for the single cell RN-seq, this technology was designed to quantify the RNA expression and single cell level as well as to actually correlate back with the coordinate on the image. For the multi-spec immune fluorescence, what you see here is a colorful image where each color represents one cell and in one channel. The main purpose of this technology is to profile for each single cell what's the quantity relative location and cell-cell interactions on the same image. Lastly, I want to introduce the nanostream Geomax that is a digital spatial profiling. This technology, what it does is it actually picked the region of interest as you can see circled by circles and then profile about 70 different transcripts or 50 different proteins from this region that are picked out of the whole slide. So for multi-spec new fluorescence and also DSP are the major technology that we use in our lab. And for today and based on the current release of our software, I will talk about nanostream spatial profiling of the transcripts. A zoom-in view of nanostream DSP is looking like this. As you can see, we see two colors. And then for the green color, those are the actually C45 color. And for the PENCK, those are the purple ones. So one thing about nanostream is you can actually select which kind of cells you want to segment and then only measure the expression profile in that set of cells. And usually people pick the choice of tumor versus immune from this technology. A further dig into that is when we started to analyze this kind of data in our lab, we found that even though we have the data collection, we could not really effective transform that into knowledge discovery. For example, which genes are actually differentially expressed between this part of the slides and this part of slides in terms of tumor heterogeneity. That is the major motivation where we actually designed this software to actually be open source and also try to solve and fill this gap between the collection of nanostream data and also try to prioritize some candidates for downstream experiment in the lab. We anticipated this kind of software might be useful for clinicians and biologists to actually really explore their data. And also it will be useful for bioinformaticians to perform a data analysis within our studio. Lastly, we hope some of the core facilities including ourselves can also make use of this data and also this software to quickly QC and process some of the data for the result. So for those that have not seen in SPA before, this is the data frame that has input for the software. This looks very similar to the nanostream count data. However, it has two key elements that is very different and makes it distinctive from the traditional quantification. The first one I want to point out is the segment selection. As I pointed before, different colors are two color tones and it's possible to add more markers or more colors. But you can actually pick one color that means immune cells or tumor cells and quantify expression only within that set of cells. That is called a segment. Second one is you can quantify the gene expression based on the region of interest. And usually what we pick is we actually visualize this region along the tumor border or inside and decide which kind of territory structure we are interested in. And then we would manually actually annotate this and select those for the machine for quantification. So those are the two elements that our software is trying to take care of. A typical workflow in our software, including QC, normalization and statistical analysis. Because of interest of time, I will not go detail into about that. But I do want to point out three major features. The first one is it's important to select which normalization method is best suitable for data. And second one is it's important to select which segment that can be used for your research question. For example, if you are looking for genes that only express in tumor cells and then you would want to actually specify only quantify tumor segment for your research. The lastly is because of this complex design from the spatial distribution, we actually implement the linear mixed effect model to take care of the patient to patient variation when multiple scans are actually taken care of. So here are some highlights features from our software. First one is we want to utilize the spatial distribution of the software which we actually allow people to visualize specifically on top of each RRI. We actually create the size of the bubble that represents the expression of specific genes. So we got this inspired by the COVID research. So we think we could actually apply the same concept on the pathology. The next thing is for the statistics, what can we use this data to answer questions? There are two kind of major questions we have seen can be answered by this kind of data. So the first one is within the same image, what's the difference between different compartment of this image? Because you can select the RRI, the region of interest, based on research question. For example, if this is tumor and this is normal, you can actually easily compare what's the difference between tumor and normal within the same patient. The second one is you can also compare different groups. For example, if this is responders, this is non-responders, and you collect multiple RRIs or multiple scans, it's possible to ask questions such as which kind of biomarker is different between these two patient groups accounting for tumor heterogeneity. So just a preview of these kind of functions in our software, we try to modernize each step into specific functions as you listed here. And the last two has not been available based on current release, but we hope to actually update it very soon. And how to start using our software? We have published our software on GitHub. It's open source, so everybody can get access. We also developed a shiny application because we value the visualization and interactive interpolation is very important for the spatial data. And we also was inspired by IC, which is very good software for single cell spatial mix. Lastly, I want to point out for future development, we're interested in adding multi-segment selection for DSP modules, and also we're interested in adding the vector, multi-spectral immune fluorescence analysis to our software. So with that, I want to thank my collaborators. Raj is really the lead developer for this software that has contributed critical amount of time and effort to make it work. Jason, as well as Tulia, both provided us valuable samples and generate data for us. Tulia's group at Flowcore generated all the data for this experiment. And we also thank our colleagues that helped test and validate our software. So with that, I also want to make an advertisement. We are hiring. So if anybody is interested in our work or in our lab, please feel free to reach out to me. So before I move on, I hope I still have some minutes to demonstrate our software. You have two. You have two minutes. Two minutes. Got it. I want to show everybody that if you go to this GitHub repository, you will be able to see our source code here. And also for the shiny application, I will demonstrate what each function looks like on the GUI interface. So give me a second. So I think we need to be ending up here. Okay. Kind of running a little bit late. So. Oh, yes. Can I just have one minute, though? So you've got your GitHub, but you could also type things in the chat window if you're... Oh, okay. Okay, that is good. That is no problem. Okay. Thank you very much for your presentation. Thank you.