 Well, I'm glad for this chance to present on behalf of our analysis working group. Kim Rathmell is my co-chair. She's currently on sabbatical in Switzerland. Chromophobienosal carcinoma, or CHRCC, represents about 5% of cancers arising from the kidney nephrin. Due in part to its relative rarity, this disease has been understudied at the molecular level, but now it's been comprehensively profiled by TCJ as the first of its rare tumor projects. Shown here is the data summary for the project, and all we have 66 tumors that were subjected to the standard multi-platform analyses as part of TCGA, and while this may not represent as many cases as for some of the other projects, we were able to move in other dimensions through data integration. So for one thing, we have here 50 cases with whole genome sequencing, which is a lot for this type of project, and so we focus a great deal of effort on mining these data. We also have 61 cases with mitochondrial genome sequencing. This would not be part of the standard TCGA pipeline. Here we use long-range PCR to amplify the mitochondrial DNA component, and we found the results from this compared quite well with those from whole genome analysis. So we can start with somatic alterations by copy and whole exome analysis. And on the right of this slide, we have a heat map of DNA copy alterations for our 66 chromophobe cases, or CHRCC, and we find the patterns for CHRCC differ quite a bit from those associated with clear cell renal cell carcinoma, or CCRCC, which is represented on the left. In CHRCC, there is whole chromosome loss for 1, 2, 6, 10, 13, and 17 in the majority of cases. However, we could find no significant focal regions of copy alteration in CHRCC by GISTIC analysis. We carried out whole exome sequencing in CHRCC to identify somatically mutated genes. TP53 was the most frequently mutated at 30%, followed by P10 at 9%. And then we had this long list of genes mutated in just one or two cases, which we understand to be relevant in terms of our biological knowledge. Now while the whole exome data is certainly an important component, it was also clear to us at this point in the project that there might be more to the story than what will be represented here. As we can move on to DNA methylation and RNA expression, and at the DNA methylation level, we observed widespread differences between CHRCC and CCRCC, involving tens of thousands of loci. And so clear cell and chromophobe are both kidney cancer, but they look very different at the molecular level. This is from the perspective of using various data platforms. Now we might simply note these differences as a finding and then move on, or we might press a bit further and ask just why are these two cancers so different. So in principle, some of these molecular differences can involve cancer-relevant pathways, but other differences may also reflect the cell of origin of the cancer, if in fact these two cancers do have different cells of origin. And this question as to cell of origin of kidney cancer is something that has received much debate in the research community, and the current theory held by some is as follows. Shown here is a diagram of the kidney nephrin. This is just part of the kidney, but it's a part of the process that's urine from the blood. And for this talk, we can think of there being two basic parts of this nephrin. We have the proximal end on the left and the distal end on the right. CHRCC has been postulated to arise from intercalated cells in the distal convoluted tubular of the kidney nephrin. While CCRCC could arise from cells in the proximal convoluted tubule. And this theory is based on previous work focused on specific markers by menohistochemistry, but we can also look at this question using a global analysis of gene expression. And so to examine this question, we needed an orthogonal data set, something outside of TCJ and even outside of cancer. We are very fortunate to have this gene expression atlas of the kidney nephrin in the public domain. This is a previously published mRNA profiling data set of micro dissected regions of the nephrin as indicated in this graphic. These are from both mouse and human specimens. And what we found with this data set is that CHRCC versus CCRCC expression differences reflect differences between distal versus proximal nephrin respectively. Here we formed a supervised analysis where each TCGA chromophobe or clear cell kidney tumor expression profile was compared via a global correlation analysis to that of each sample in the kidney nephrin atlas. And so the rows of this heat map represent kidney nephrin sections and the columns represent kidney cancers. Here we see that CHRCC clearly shows high mRNA expression correlations with the distal regions of the nephrin, while CCRCC was highly correlated with the patterns associated with the proximal nephrin. This is seen in both mouse and human data sets as a very striking pattern clearly distinguishes these two cancers as being distinct from each other. And this will be in line with the expectations of many as well as helping to put all these data into some meaningful context. Next, we will cover mitochondrial DNA alterations. Now, while we did not find many nuclear genes being frequently mutated other than TP53, mitochondrial DNA or MTDNA is typically more vulnerable to mutation in cancer and encodes for some important genes. And so we sequenced the mitochondrial genome for our CHRCC cases, represented here as a lollipop plot of somatic events with greater than 50% heteroplasmy. And so most MTDNA copies in the tumor cell would have the mutation. MTDNA is circular, about half the mutations detected were in the D-loop region, which is non-coding. But MTDNA encodes for 13 proteins involved in respiration and oxidative phosphorylation, many of which are found mutated in our data. Here are the MTDNA encoded genes in the context of their associated pathways. These genes, now highlighted, all involve the electron transport chain. And most of the mutations in yellow occur in complex one, in particular in the ND5 gene. And due to the type of mutations represented here, we think these are likely to result in a loss of complex one activity. Also represented on this slide are the expression patterns of the nuclear encoded genes, with red meaning high expression in CHRCC versus normal, and blue meaning low expression. And what we should take note of here is that there is a lot of red in this picture, with some blue at key entry points into Krebs. And so while many cancers, such as CCRCC, would seek to down-regulate Krebs cycle and oxidative phosphorylation, showing a type of Warburg effect, CHRCC, if anything, shows an anti-Warburg effect. This is a cancer that relies upon oxidative phosphorylation. And other pieces of data would support this as well. Now it's interesting in that some have hypothesized that the MTDNA mutations should lead to a Warburg effect by effectively knocking out complex one. However, this would not be reflected in our data on CHRCC. And all of this would seem to suggest alternative roles for MTDNA mutations in cancers, such as CHRCC, that rely on oxidative phosphorylation. So finally, we'll cover some key findings from whole genome analysis. From our whole genome sequencing data, cotidius could be observed in some of our CHRCC cases. Cotidius is this phenomenon involving highly localized substitution mutations, which has recently been observed in other cancer types. And consistent with previous observations, we found that the regions of cotidius and CHRCC were often found in the vicinity of genomic rearrangements. Now it's interesting in that this pattern shows up in only a fraction of our cases. And this would invite a comparison to be made using our other data platforms. And so we compare gene expression profiles between CHRCC cases with and without a strong cotidius pattern. And we identified 29 differentiated express genes with an FDR of 5%. What's interesting is that this list of genes included TERT at the very top, which had the highest level of expression in the cotidius cases. TERT gene itself showed a wide range of expression levels across the HRCC from undetectable to hundreds of units by RNA-seq. And this caused us to ask, just what might be driving the expression of this gene in some of our cases? And Caleb Davis examined DNA copy number within the genomic region surrounding TERT. And while we did find some copy variation from normal for some cases, it was not at levels that would account for the extent of deregulated expression. However, what Caleb noticed was that multiple cases show these abrupt changes in copy number at points that fell within the region 10 kb upstream of the TERT transcription start site. What this suggested was the existence of structural breakpoints resulting from genomic rearrangements. And so working with Peter Pars group, we were able to map out the coordinates of seven structural variants associated with TERT promoter region by WGS analysis involving six cases. Each of these entails two fragments of DNA from different regions of the genome becoming fused together. What's notable is that these TERT promoter-associated structural variants, or SVs, correlate with high TERT expression. The six cases having the highest levels of TERT were the same six cases for which we found a TERT promoter SV. In most of the other CHRCC cases, TERT levels were undetectable. Three cases did have the previously described CT2-AT activity and promoter mutation, these shown only moderate levels of expression. And so this would appear to represent a novel mechanism for deregulation of TERT whereby the promoter region is disrupted by genomic rearrangement, possibly by interfering with the activity of a key repressor. We carried out validation of these TERT promoter-associated SVs. Here's an example case of a structural rearrangement involving cancer KN8-435, where inversion is involved. For each breakpoint, we designed PCR primers to span each of the two genomic regions involved, as highlighted here. And by combining these two sets of primers, we could amplify the breakpoint region in the tumor sample but not in the corresponding normal sample. And then we subjected this PCR product to sequencing for a final confirmation. So of the seven rearrangements originally identified by WGS, we could independently confirm six, involving six cases, by PCR. And this would represent a major funding for our project. And it leads us to ask whether TERT may be similarly deregulated in other cancers. So in conclusion, we've carried out comprehensive molecular analysis of a rare cancer subtype as a platform for discovery. We find that global molecular patterns may provide clues as to a cancer cell of origin. MTDNA sequencing has been incorporated into multi-platform electric characterization of a cancer. And finally, we have this important discovery of recurrent genomic rearrangements involving TERT promoter region. That brings us to the end. I'd like to thank all of my colleagues who were involved in this project and made substantial contributions and without whom this work would not be possible. And also, I'd like to thank Kim Rathmill, my co-chair. And I thank you for your time. We'd be happy to take questions. We have time for a couple of questions. Hey, Chad, great talk. I had a couple of questions about the TERT rearrangements. First was how the expression levels compare to the expression levels in other tumor types with frequent promoter point mutations like gluoblastoma and melanoma in cases where you have those rearrangements. And second is what you actually see about telomere length in the samples that have undergone whole genome sequencing that have the promoter rearrangements. Those sounds like great questions. I think in terms of, well, we've looked at the TERT promoter mutations in our chromophobe cases. I don't know, that's generally not a standard thing that's checked into the other TCJ data sets, which have all been done on the same platform. So I think that's something we can look into. My understanding from literature is that the TERT promoter mutation should increase the expression by two to four-fold. I think what we saw with the six cases that we had with rearrangements, it was hundreds of units much higher than what we saw for the three cases that did have the promoter mutation. So I don't know if there's enough data in the TCJ to do a comparison, but I think that's something we need to look into. In terms of telomere length, I think that's also a good question. I guess I'd have to look into how we can get that information. Was there any overlap between the two, like clear cell and chromophobe in your data set? Any overlap between the two tumors? Because previously, from TCJ, there was some overlap when the batch effect occurred. It was Dr. Arkbrie's group at Anderson who did it. So I'm interested to know, was there any overlap between the two tumors at the molecular level? Yeah, that's a good question. I think in the clear cell project, I guess kind of a wrinkle is that some samples that got in that were supposed to be clear cell and turned out to be a chromophobe. And it comes out in the molecular analysis when we caught that. And we had a pathologist review the cases. Those were taken out of the clear cell paper. They are not featured in this study. And I think there's reasons for that. I think the samples that we analyzed here were dedicated samples that were known to be chromophobe. And I think we had the pathologist look things over. And that was consistent with the original diagnosis. Thanks. Yeah. Did you say 6 out of 7 validated of the SVs that you found? Did I hear that right? Yeah, there are seven structural rearrangements. One case had two structural variants in that 10KB region upstream. So that was a little more complicated case. So I think we could validate one of the two. We couldn't validate the other one. But we think it's real from the analysis. I think that this case is a little bit more missier. It's got more rearrangements going on. But it looks as if what we're seeing has kind of been confirmed. Yeah. Any more questions? Thanks a lot, Chad. OK. Our next speaker will be Bulent Axoy from Memorial Sloan Kettering. I probably just butchered your name, sorry. He'll be talking about prediction of individualized therapeutic vulnerabilities in cancer from genomic profiles. Thank you.