 Thanks for this great opportunity for us to present our work here. And I'm going to talk about a new feature that we develop in the CBIO Cancer Genomics Portal for visualizing and analyzing clinical and genomic data for individual patients. CBIO Cancer Genomics Portal is a web resource that is open access and for interactive exploration of high-dimensional cancer genomics data. Our goal is to lower the barriers between cancer genomics data and the cancer researchers, including well-lapped biologists and clinical researchers and computational biologists. With this portal, one can very quickly visualize and analyze genomic alterations in an iterative and interactive way. And we hope that with this portal, we can really help the cancer researchers to translate the complex genomics data to biological insights and clinical applications. I would like to mention one key abstraction, namely, altered gene. We consider a gene as altered if it is mutated or amplified or deleted. So this simplifying concept actually is very powerful and enables users to derive hypothesis regarding frequently altered gene set pathways. And we just released this patient review. We aggregate integrate diverse data sets and filter them and present them to the user. And we want to do something more interesting and more intelligent because tumor samples, oftentimes they have hundreds or even thousands of alterations, which of them are relevant to the disease. And can we use this relevant event to define the choice of treatment? And actually, experts, they are pretty good at intuitively identify them. Can we automate part of this process and can we even improve it? So our goal is to develop an interactive system to facilitate personalized medicine research for cancer and ultimately guide a patient treatment. So in this patient-centric view, we integrate diverse data, including clinical data and genomics data. So the clinical data includes patient information, disease status, and pathological reports. And we are going to support tumor images and treatment data as well. And the main part in the patient view is the molecular profiles. Basically, we support mutations and copy-num alterations. And gene fusion and expression will be supported later. After we get those alterations, next step, we are adding functional and statistical annotation to those alterations, including predicted mutation impact by mutation assessor and cancer gene annotations from Sanger cancer gene and COSMIC. And COSMIC and cohort context information, such as significantly mutated genes or copy-num altered peaks, we also bring in drug data information and clinical trials from resources such as NCI drug knowledge base. And after we're putting this together, we present this to cancer researchers and clinicians in an interactive way. And we are also building a wiki system that can be used for the user for annotated specific tumors. Enough said, this is an example of our released product of the patient view in the current portal. In this particular case, this woman is a 69-year-old serious ovarian patient with stage three and grade three. And we can get more clinical data from here. And this summary view is basically summarized the mutation event and copy number events. I'm going to take a few minutes to explain the detail and hopefully make this not only colorful but also meaningful to you. This graph provides copy number events and mutation events across the chromosomes. The copy number check provides the copy number segments. And from here, we can see that actually, in this patient, the copy number is pretty much altered and this same characteristic was observed in serious ovarian cancer. And in the mutation check, we plot the histogram of number of mutations across chromosomes. This is a sketch plot of number of mutations and the fraction of copy number altered genome. Each dot represents a tumor sample in the cohort study. And the red one is the patient we are currently viewing. In this panel, we list the mutation events in this patient. In this patient, there are totally 35 mutations and four of them are reported as interesting by the portal based on recurrence cosmic or annotated cancer genes. The first mutation is a TP53 nonsense mutation. TP53 mutation is almost universal in serious endometroid. The second mutation is a phosphatase 2 missense mutation. It is a recurrent mutation. There are 26 other samples in this study have mutated phosphophase 2. Three of them have this mutation on the same amino acid. The music also showing that it is a significantly mutated gene or recurrent mutated gene. And it has also overlap with the cosmic data. Mutation assessor predict this mutation to have a high functional impact. And from the 3D structure through the mutation assessor, clearly it is a mutation in the highly conserved region. In this panel, we list the copy number events. And there are totally 385 copy number mutated genes. And 17 of them are reported as interesting. For example, the second U1 is reported because it's a recurrent gene based on gestic. And it is actually a focale amplification. But for example, this EGFI actually is reported because of annotated cancer gene. And it's a rare event. It's not recurred and not in gestic peaks. But it could be interesting. And EGFI is a well known drug target. For example, for this drug icon, we can get the drug information with clinical child information linked. The second and third tab, we list off the mutation and copy number alterations. And the fourth tab, we also link to the pathology report for this patient if available. And from this patient view, we can link back to the cancer study view. The cancer study view presents off the clinical and cancer genomics data in an iterative way. For example, for the clinical data, we plot off the histogram and pie charts. And they are all interactive and can be cross filtered. For example, for this histology plot, if we click on the serous pie, then off the serous cases in this plot will be selected. And immediately we can observe that this patient is an outlier because it has a very low copy number alteration and has a lot of mutations, unlike other serous cases. And if we click into this patient and have a look, clearly this patient almost has no copy number alteration and has a lot of mutation events. And it has such as a K-RAS, et cetera, which are characteristic to serous cases. Therefore make this case very suspicious. And back to the cancer study page, we also list the significantly mutated genes and copy number peaks. Next I will do a little bit demo in the next two minutes. I hope the internet is working. Okay, so this is the front page of the query page. And for the time sake, let's select the endometria case for demo. And we leave this mutation, copy number change as default. And we can select some genes from a music list. For example, this beta-container and K-RAS and summit. This will bring us to a summary page, including a so-called uncle print. From this uncle print, it is clear that there are a lot of mutation events occurring in these two genes represented by this green dot. And let's look at those mutations. So here in the mutation diagram, most of these beta-container mutations occur in the unterminal area. And they have well-known functional impact, actually. If we sort this, we can have, it's a very long list in the unterminal. And similarly for the K-RAS, most of them actually, 35 of them actually occurred on the G12 position. And if we sort, it's a long list of them, and they have well-known functional impact. Getting back to the summary page, there are six amplified, six cases with amplified K-RAS. And if we are interested in whether this amplification is functional or not, we can have a look at the plot. Let's select K-RAS. And this is a plot between gene expression versus gestic, and this rightmost, rightmost plot. In this, from this rightmost plot, we can clearly see that there's at least one sample with a very high expression level of K-RAS. And from this uncle print, we can kind of spot a mutual exclusivity pattern. And let's have a look at the p-value here as 0.02, which is significant. So the next question is why we still have this overlap, still have cases that have both mutated in both genes. And we can take a look at maybe the first one. This patient has a lot of mutations, more than 6,000 mutations. It could take a while to load it. But I think that this is exactly the reason why this patient has both mutations in both genes because of the higher mutation account background. Okay, I will give up on this one. And I have a preloaded version here. So this patient has a functional mutation, K-RAS mutation, and beta-catenal mutation is not a well-known function here. Okay, for the time sake. Okay, I will maybe show a little bit on the patient, on the cancer study view. All right, this is loaded. And the cancer study view actually plots off the attributes and the scatter plot. And the patient is remembered. And if we clear, then we clear all. And we can select off the living patient with recurrence and et cetera. With that, I would like to stop here because of time. And we are going to have a workshop at the Alexandria Room. And with that, I would like to thank my colleagues, especially Nikki Schwartz. He's been managing this project and coming up with a lot of ideas to the patient view. And it is a joint effort of the whole team, including Ben Amon, owner Gideon Boris, Ethan and Chris. Dr. Lavigne from MSKCC provides a lot of feedback and his group provides a lot of feedback to the portal and to the patient view. Thank you all. Congratulations on the live demo. Thank you. So what's the status of your clinical interface, you know, physicians using this, you're gliding this all into some position where the less academic condition can use the tool? Yeah, I mean, for now, basically, it's for researchers, clinical research, like cancer research, basic cancer research, but we're hoping that it really can be used in clinical setting. Thank you. Thank you. So I have a question when you're deriving the individual copy number alterations. Do you use the cohort context at all, or are you just looking at tumor versus normal for the same patient? The... You mean... So when you're deriving the copy number alterations for a particular patient, do you just look at that particular patient's tumor data and normal data, or do you also include the cohort context like just take information? Yeah, yeah, we use the just take information, actually. Okay. Yeah. Okay. Two more questions. Quick question. Subha Madhavan, Georgetown University, fantastic presentation, thank you. I was wondering how much pre-processing occurs in the back end prior to loading the data into the CBI portal. I mean, for all the 7000 or so samples that you have on your portal, I can imagine, you know, it takes quite a bit of time for the various data types that you're presenting here. So can you shed a little bit of light on what kind of pre-processing that you do before loading the data into the CBI portal? Yeah, actually, most of the data for the TCGA, we get it from like FIHOS, and we have some pre-processing, for example, for the expression data, and the mutation data, we add a lot of annotation to the map file. And for the copy number data, we also have some pre-processing. Yeah, it's a lot, and we have people like in charging of managing those data and pre-processing, and we have a whole pre-processing module. Quick follow-up. Do you have a timeline for how long it takes for a particular study, let's say, 100 samples with three different data types? Timeline, you mean? Yeah, for pre-processing and getting it into CBIO. How long does it take to get... You mean the... About 100 sample study into the CBIO portal. Okay, it may be like a few... One hundred, maybe less than one hour. Okay, thank you. Last question. Yeah, I just have a quick comment, maybe also for the community, so that when many people think, if it's apparently love something, many people think about like bioinformatics software and tools as a black box. And it's appreciated that you talk about the workflow, how to use the black box, but I think that like more insight, more explanation to the life sentence, like what's inside the black box, what's the computational approach, the limitations, why it can... What kind of work it can do, what kind of work it cannot do, so that the inside the black box, the more inside to the computational approach, so there will be more helpful, I hope, to say more information like during workshop. Thank you. Okay.