 Good morning. Today I'll talk about metal exclusivity between genomic alteration occurring across multiple cancer types We know We know that in spite of Largest heterogeneity recurrent alteration frequently Target few specific and hygienic pathways and we also observed and recently showed That functional alteration targeting the same pathway frequently occurred in a mutually exclusive manner and we have here a nice example Coming from the latest DCGA paper In which we show that inactivation of beer say one and beer say two Almost never occur So to systematically identified Jeans that are frequently outred in a mutual exclusive way We develop a method called MIMO that stands for mutual exclusivity models MIMO integrates semantic mutation copy number alteration and allows the user to specify a set of alteration for one or more genes For example to include beer say one hyper mutilation After the frequently altered gene has been detected We connect them using a background network all the fully connected gene sets are when extracted and Alteration affecting each gene set or module are tested for mutual exclusivity Only models whose set of alteration show a statistically significant trend towards missile exclusivity are then pulled out from a method as a result MIMO has been successfully applied to several DCGA cancer project And I'm focusing today on five of them including updated data set for the series of adding cancer and glioblastoma deformer as well as more recent project Like the colorectal adenocarcinoma focusing on the non hyper mutators subtype the uterine corpus endometriot carcinoma focusing again on the non hyper mutators and no serious subtypes and I will talk a bit more on invasive breast cancer that among these five is the one for which we have a Larger the largest data set with full complete genomic profiling in each of these cancer MIMO identified Mutually exclusive pattern of alteration affecting several oncogenic pathways including RB signaling P53 signaling DNA repair and PF3K aka T signal escape, but today I will focus specifically on this on this PF3K aka T pathway including RTKs and RAS This pathway is frequently altered across all these cancer types and more interestingly by different mechanisms I want to give you a first overview of a set of targets that have been identified by the method and You can see how high percentages of tumors are altered in each cancer type and It's quite interesting seeing the pattern of mutil exclusivity that characterize alteration affecting this genes Sometimes this pattern are cleaner like for endometrioid or colorectal where we have few targets highly With altered with high frequency some other times in like in GBM or breast where we observe a larger heterogeneity We also observe some overlapping cases But what probably you may have you may have already noticed is that one of the cancer types that we introduced before is missing here Meaning ovarian cancer indeed. We did not find a strong signal coming from Frequently altered genes in serious ovarian scan in serious ovarian cancer So we ask ourselves are there low frequency but functional events affecting the PF3K aka T signaling cascade in ovarian cancer And when we look at the core components of this pathway We actually see that there are multiple low frequency event that target the PF3K aka T signaling including RTK simplification a rare but functional mutation in rust and rough P10 down regulation. I'm up to K4 down regulation When we aggregate all this Alteration we can now sum up up to 24 percent of ovarian cancer tumors that have alteration in this pathway and all these alteration show also a nice pattern of misal exclusivity even for we are looking here at right event and In conclusion, we can say that this pathway is actually altered in all the five cancer types that we analyzed Sorry, can I go back? Yeah, I'd like to talk a bit more on Invasive breast cancer as as I said it before we observe here the largest heterogeneity of alteration We saw somehow expect that we know that breast cancer is a very heterogeneous disease But can actually be thought at least as two distinct diseases And I'm referring here to a distinction between basal tumors and non-basal tumors Did this distinction originally identified from RNA expression signature? Apparent is Clear by looking at whatever kind of genomic or epigenetic alteration in breast cancer so we ask if This pathway is altered preferentially one of the two sub types or by different mechanisms in the two subtype When we first broke down the set of alteration that we found We see that up to 74 percent of non-basal tumors have alteration on at least one of the gene From the pathway show here on the left while only 24 percent of basal tumors have so we ask Is the pf3k pathway altered by other means in basal tumors and to answer this question we want looking at a better than pattern of mRNA expression meaning upper regulated genes or downregulated genes independently of copy number and mutation status and the first evident result that we have was that p-tan is frequently downregulated in basal tumors independently of copy number and mutation status you can see in this plot p-tan expression versus copy number status and all the yellow dots are basal tumors You can see that most of the downregulated samples are indeed basal tumors When we went looking at the AKT level that the AKT phosphorylation level for this tumor We actually see that all the p-tan downregulated samples show on average higher AKT phosphorylation We suggest that this event is functional and p-tan downregulated samples actually Contribute to activate the AKT signal p-tan was not the only gene that we found with having an abundant mRNA expression pattern and Indeed we found a very interesting pattern for AKT free expression in basal tumors AKT free expression in basal as a strong by model distribution and we can see that up of 30 percent of basal tumors have AKT free over expression When we put together this observation We actually see that we now have 50 percent of basal tumors that have alteration in the PA3K AKT signaling cascade and alteration in AKT over expression of AKT free is nicely mutil exclusive with downregulation of p-tan in these tumors So I'd like to give you a summary of a overall extent of alteration in the whole set of genes that we identified from our mutil exclusivity analysis and You can see from his heat map each column is one of a cancer type that we analyzed You can see appreciate some similar pattern like p-tan, p-free CA, R2 or p-free R1 But they're commonly altered across all cancer types as well as some specific cancer-specific alteration like IGF2 over expression in colorectal or EJFR amplification in glib-blastoma In conclusion I'd like to stress once again that MIMO systematically identified mutil exclusive alteration Target in oncogenic pathway across multiple cancer types In particular today, I focused on the PA3K AKT signaling which is consistently altered in cancer and more even more interestingly by a larger heterogeneity of mechanisms Finally, I think it's important to stress that all this kind of analysis Across these cancer types have now the ability to unveil The underlying heterogeneity of the disease was suggesting candidate therapeutic targets in each single subtype And with this I would like to Thank all the people that collaborated with me in particular Sorry particular Chris, Nicky and Ethan who collaborated with me in developing MIMO as well as analyzing all the TCGA data sets Thank you. Thank you Giovanni questions great talk one question is that with regards to Determining significant mutual exclusivity just from our observations of Relatively, let's say rare somatic mutation Mutations or somatically mutated genes and tumors that it's pretty easy to find Let's say a set of five or six genes that show that nice staircase pattern that implies mutual exclusivity So what are the steps that you've taken whether it's permutation analysis or some analytical statistical tests that You're correct for if you look if you're looking at rare events It's easy to find this Missile exclusive pot and in our case we test we use a permutation test That is basically a Monte Carlo Markov chain permutation strategy that preserve the number of alteration per genes and the number of alteration per samples So in this way, we believe we have an accurate way, we are able to preserve a Cancer specificity in altering oncogen and tumor suppressor as well at the heterogeneity show by different samples Giovanni that was a beautiful talk. I have a quick question. Could you speculate a little bit of what you think is Might turn out to be the basis for the different mechanisms of pathway alteration and the different types of cancer Very broad varies. I'm asking you to speculate here honestly, I Think it's a pretty common feature to serve this large heterogeneity in a lot of cancer show different mechanism alternative mechanism but we know this alteration most of the time emerge randomly during But it's it's the process is and are selected and if for the Advantage that may give to the cancer cell to proliferate and altering the key pathways I guess where we still have to figure out what? what causes the Different mechanisms to be pursued for the same pathway and different types of cancer. It'll be an interesting question to sort out Okay. Thank you very much great talk