Skip to content

2026-04-22-precision-for-all-multi-omic-technologies-inequalities

Date: 2026-04-22 Primary topic: single-cell-spatial-omics Cross-referenced from: single-cell-spatial-omics Word count: ~13,269

Auto-generated captions

This transcript is Vimeo auto-generated English. Expect misheard names and technical terms. Use the ripgrep/search approach for exact quotes.

All right. Thank you all for joining this session. I know it's the end of the conference, and you probably could have caught an earlier flight, so we're grateful. We were worried it would just be the three of us having a chat to each other . So we're very grateful that you've made time to join, and those online as well. I think there are a few people online. And obviously, these sessions are recorded, so I think it'll be great. Whatever discussion we have here will obviously continue on beyond this room. So my name is Yaw Bediako. I am an immunologist by training, and also more recently, a biotech co-founder, running a genomics biotech company based in Ghana. And I'm very grateful for the opportunity to put this session together with two of my colleagues and my friends, actually, where we're going to talk about what I think is a really important and topical issue, relating not just to representation, but really precision medicine for all. It says, "Precision for All: Harnessing Multi-Omic Technologies to Eliminate Cancer Inequalities." And I like the idea where it's focusing on for all. It's not just about eliminating inequalities, but what does it really mean when we say access for everyone? There were really good talks yesterday on ancestry-related disparity, and Atinasha gave a great talk, and I think this is a great continuation on that topic. Beginning to think about it's one thing to study underrepresented groups and learn more about cancer in those groups, but then what do we do with that information, and how do we make sure that that information actually feeds into improving access to better healthcare? Obviously, the end goal is to make sure that anyone who gets cancer, no matter who they are or where they are, gets great access to care. But obviously, there's often a huge gap between understanding and implementation and access, and I hope that we can unpack a few of those topics today. So the three of us are each giving 20-minute talks. I think what we've decided to do is something that seemed to work well in a few sessions I attended yesterday, is we'll hold questions to the end so we can have a holistic discussion. Our talks are designed to build upon one another, so hopefully the questions you may have will be best served or best addressed at the end when a few people can chip in. So looking forward to having a good discussion at the end of the three talks. Since I decided to start off so I could set the stage for them and then sit back and listen to their talks, so I will begin. I thought it was good to just realize that precision medicine is not that young, and I thought this is a really good slide that kind of charts, in some ways, the history of precision medicine, beginning with Herceptin in 1998. And really the point I'm trying to make here is one I've just made, that we've known a lot, the number of precision therapies that are almost cliches in some places, but we still have a significant gap when it comes to especially African ancestry individuals. So over 30 years of precision medicine, in 2012, HG Africa, which was an NIH project to understand hereditary and genomics in Africa, was launched, and did a number of great things, including a cover of "Nature" in 2020, where for the first time, 400 people from Africa were sequenced. A whole genome sequencing was done and three million new variants were discovered and it was like, wow, how diverse Africa is. And yet still now in 2025 or 2026, less than 3% of available genomes are Black, and most of those genomes are actually African American, which are an admixed population. So it's actually less than 1% of genomes are from people living on the continent of Africa. So I think that's a great place to set the baseline for today's discussion, is when we say precision access for all, are we thinking all people in the United States, or are we thinking broader? And my encouragement would be, let's think beyond just the US, beyond Europe, thinking about parts of Asia and certainly the continent of Africa, where you have significant burdens of cancer, and yet you have significant gaps in access to even basic older drugs. And what does that mean as we're in an age of immunotherapies and new kinds of treatments? So I'm going to talk to you today about what we call the African Cancer Atlas. It's really meant to be a response to TCGA. Couldn't quite figure out a way to get the G in there, so we stuck with TACA, T-A-C-A. But this is an initiative that was launched by my company, where Yamachi is a small, five-and-a-half-year-old startup headquartered in Accra in Ghana. And really our interest is how do we leverage Africa's genetic diversity to improve how we develop drugs for cancer for African people, but more importantly, for everyone? Because Africa is where we all come from if you go back far enough, and Africa has unparalleled genetic diversity that could serve drug discovery for a variety of diseases, but especially cancer among all people. And so that's what Yamachi is trying to do, is to try to unlock the diversity of the African continent in the service of better, more inclusive drug discovery. I have no financial relationships to disclose. So I've already preempted this slide by showing you the two slides ago. Africa has largely been excluded from genomic research and even more excluded from cancer research.So Africa accounts for 17% of the world's population, and yet less than 2% of available genomic data in cancer research is from Africa. You can see from this figure that was published in 2018, that when you look at cell lines, when you look at genomic research, when you think about clinical trial, in fact, every stage in the development pathway, people of African descent are significantly underrepresented. But I think it's quite striking that the greatest underrepresentation is actually in genomic data, which in many ways is the beginning of a lot of the research that we've been seeing talked about over the last four or five days. And so if you're underrepresented in the original training sets, then it's no wonder that you'll continue to be underrepresented as you go down the value chain. Now, this underrepresentation flies in the face of the fact that Africa is home to the world's most genetically diverse population. Our species evolved out of the continent of Africa. We spent 99% of our evolutionary history on the continent of Africa. And this is that study that made the cover of "Nature" in 2020, where they basically identified or did whole genome sequencing on about 400 individuals from across the African continent, and just really maybe for the first time, or maybe the clearest demonstration of just how diverse the continent of Africa is. So on the figure on your left, you're looking at variant sites per genome, and this is layering it with 1,000 genome data and other data from additional ancestries. The African ancestry populations are the one in that step that goes up. And you can just see the fold increase in genetic variants for the African populations relative to the European populations. And you can also see that from the PCA, you can see how even within Africa, there's significant diversity. Okay. Is that better? Okay. So even within the continent of Africa, you can see significant diversity from West, Central, Southern, and East Africa. So Africa is very diverse. Within Africa, we're very different. In fact, the colloquial saying is we're more different from one another than we are from anybody outside of Africa. And I think this figure kind of shows that quite clearly. So you could ask yourself, when we started sequencing, why didn't we just take sequences to some of Africa's biggest cities and start sequencing African people? We'd probably be a lot further on in understanding human genomics than what we currently are. And when you think about African genomes, and Nyasha is probably better equipped to talk about this than me. Linkage disequilibrium blocks within African populations are shorter, significantly shorter than you find in other ancestry populations. And so this presents opportunities, really to identify causal variants. It's much easier to identify causal variants in African populations than in others because of this genomic architecture. So I'm going to present a few papers that have come out relatively recently describing some of the clinical implications or the molecular implications of genetic diversity. The first author of this paper is sitting in the room, Seto. So if you want to... Yeah, he's waving. So, if you have any questions about this slide, talk to him, not me. But this was a study where we really did a meta-analysis on two large cohorts, looking at genetic diversity, mostly in this case in sequencing panel data, and really demonstrated that for certain mutations like TP53, they appear to be enriched for a variety of cancer types in African populations. Whereas for other genes like TERT, you had what seemed to be depletions in the African population. Now you could ask, is this just based on the way the panels are designed? So these panels are obviously largely designed based on European ancestry populations. But it certainly raises the point that you probably need to think about your panel design when you're looking at more diverse populations since you do have these differences in variant distribution in different ancestry populations. This is a study that we did ourselves at Yamachi, really a meta-analysis, looking at PharmGKB data, so basically pharmacogenomic data, for two very cheap and readily available chemotherapy drugs, cyclophosphamide and capecitabine. And the simple question we asked was, if we look at existing published data and we look for specific alleles associated with either toxicity or better or worse efficacy of these very common, very cheap, highly used drugs across Africa, do we see differences when it comes to genetic ancestry? And the short answer is yes. For a large number of these alleles known to be either mediators of toxicity or reduced efficacy, we found that there was difference in their allele frequencies between people of Black African ancestry versus Asian, Caucasian, and Hispanic individuals. So you could even argue that for even our old chemo drugs that, let's face it, are the most available in places like Africa, certain genetic screening could be useful to improve the efficacy. So we call this precision medicine on a budget. So if you can't have access to immunotherapy, but you have access to old chemo, within Africa, given our diversity, you could do it on some of these tests and actually potentially improve the outcomes for individuals on these drugs. This is a paper with Nyasha as co-first author on. So once again, as you can see, I've carefully picked papers so I don't have to answer detailed questions about. But in this paper, what Nyasha and her collaborators did was really do a comprehensive analysis of TCGA data looking to determine whether there were differences by ancestry in somatic mutations, methylation, or gene expression.And what they found was importantly that there were subtype-specific differences. So the top panel is really describing the differences in different, in this case, breast cancer subtypes or cancer subtypes in a variety of cancer types. And as you can see, they vary by ancestry. So there are certain ancestry-associated subtypes that predominate. So you can see, for instance, in breast cancer, basal subtypes are more common or seem to be higher represented in the African ancestry individuals relative to others. And so they found that it was important to control for subtype when you are looking for ancestry-specific differences in mutations or in gene expression. And even when they controlled, they still then found ancestry-specific differences. So for instance, PIK3CA and TP53 were more associated in African than European. So were different between African and European. However, when you controlled for subtype, FBXW7 was the one that remained significant. But really, what I'm showing here are these evidence that subtype and indeed ancestry is incredibly important and has important implications for how cancer manifests and the potential mechanisms underlie cancer. This is another paper published from Melissa Davis' group, where they looked at, in this case, RNA-seq of, in this case, what they called an African-enriched population. So they had African-American individuals, but they also included individuals from continental Africa. So they had East African people from Ethiopia, West Africans from Ghana, and really looked to see how African ancestry would correspond or would affect gene expression in women with triple-negative breast cancer. And what they found was African ancestry was associated with differences in transcriptomic signatures. The heat map, basically, sorry, it's a little bit fuzzy. Not sure why it's looking like that. It wasn't like that this morning on my computer. But anyway, people with "lower African ancestry," which is on the left, clearly distinguished from people with higher African ancestry. So even at the level of gene expression in women with triple-negative breast cancer, you see two very different signatures. And when they deconvoluted these signatures, they found a number of immunological differences that were distinguishing between these groups. In particular, they found greater immune cell infiltration. So signals associated with immune cell infiltration in people with high African ancestry. But even though there was greater infiltration, they had a more immunosuppressive phenotype, suggesting that the cells that were coming in were more naive or suppressive than in European or other ancestry individuals. So once again, as you think about how we apply this, we're talking about hot tumors, cold tumors. We're talking about immunotherapy, understanding the immunological environment within which you're trying to deploy these is really important, and we cannot assume that everybody is the same. And so it's important to understand these at an ancestry level. Now, moving further into the continent. A part of what I said in the beginning was there's a need to localize some of this research. So this is a study from Kenya, looking at a small number of African Americans, Caucasian Americans, and then Kenyan individuals. And once again, looking at RNA-seq data, looking at transcriptomic data of people with triple-negative breast cancer, and clearly showing distinctions between Kenyan individuals and African Americans. So this once again makes, as I showed you earlier on, Africans are very diverse. African Americans are an admixed population. Many African Americans trace ancestry to West Africa. And so here you have African Americans compared to East Africans, and you can see significant differences. So at the beginning, I mentioned how most of the African data we have currently in genomic databases comes from African American individuals. You can see how that is not a good surrogate for individuals across the whole of Africa. So it's important that we begin to broaden the base of data that's coming in. And that is beginning to happen. This is a paper that was just published last year in "Nature Comm" for the first GWAS of continental Black African breast cancer patients. So it took us till 2025 to do the first proper GWAS of breast cancer in African women. So it's done in South Africa. And interestingly, they identified two novel loci that were actually not found in data from West African people. So this is a Southern African cohort, and you can see identifying unique loci, risk loci that would not be found in a typical West African or indeed African American ancestry population. And actually demonstrated in this paper that polygenic risk scores developed on primarily European data performed very poorly in this population. So once again, the applications of the work we do, we can't just transplant now even polygenic risk scores from populations in Europe or North America that are predominantly European to people in Africa. These scores tend to underperform. This is a study that was done in Nigeria, also quite recently. Once again, looking at breast cancer, in this case, looking at a panel of 50 cancer-related genes. So using the Illumina Amplicore Cancer Hotspot Panel. It was done in Solomon Rotimi's group in Covenant University in Nigeria. And what they managed to do was paired germline and somatic applicant-based sequencing, identify a novel germline variant in TP53. Now, I think that on its own is quite nice, but I think this also speaks to the ability to do this kind of work in Nigeria. And I think that was part of what they were proving was as we think about access to immunotherapies, we need to also have the ability to do genomic testing in these contexts locally. And so I think this was an important first sign that you could do pretty high-quality genomic testing in Nigeria.Before I get to TACA, this is the last thing I will show you. This is a study we did, currently under revision, where we did a retrospective whole exome study in Northern Nigeria. Northern Nigeria are predominantly Hausa people. Most of the Nigeria data in 1000 Genomes and in even HG Africa is Yoruba. So actually, when you think of Africa, they just use Yoruba, which is one ethnic group in Nigeria. And so I think this is probably the first time, at least that I've seen, that Hausa people have been sequenced. And I think the takeaway here, here we didn't have access to blood, so we couldn't do germline sequencing. We could only do somatic from FFP samples. So calling novel mutations, somatic mutations is a bit challenging. One of the reviewers has let us know that, so we're trying to see how. And actually, we used a Yoruba as our reference, and the reviewer called that out, which I thought was kind of cute, considering we were trying to do this for the first time. And yet, the pushback from a reviewer is that why are you using Yoruba when these are from Northern Nigeria and probably not. Perhaps the reviewer is a Nigerian, I don't know. But anyway, it is a valid point. We found a large number of seemingly novel variants. And even within cancer known genes, we found a lot of-- 67% of the 56 samples harbored novel variants in known cancer genes. So I think it just paints a picture of how little we understand about cancers in these populations. So I've tried to set the stage to tell you ancestry is important in cancer, and we're battling both the fact that we have African ancestry individuals being incredibly diverse, and we also have the least amount of data for those populations. And the truth is you can do small studies over and over again. At the end of the day, sometimes you need volume. TCGA is big for a reason. UK Biobank is big for a reason. Your N needs to be big enough to have sufficient power to understand these associations. And so a couple of years ago, we had this idea that we're not going to wait for our governments to figure out that we need to do a TCGA for Africa. We're going to try to do one ourselves. And so we called ours TACA. It's an ambitious project to ultimately generate 15,000 genomes of African ancestry individuals from really paired genomes from 7,500 people. Phase one has begun, thanks to a partnership with Roche. So we are currently funded to generate 3,000 genomes from 1,500 people. These are whole genomes, paired blood, and somatic. And this would be the largest collection of African genomes to date. So we're quite excited by that, but obviously, there's a lot more that we'd like to do. So prospective recruitment for TACA has begun. We're currently recruiting in Côte d'Ivoire, in Ghana, in Nigeria, in Kenya, and the Seychelles. That's the dot on the side. So the Seychelles islands off the east coast of Africa. And we're currently waiting for-- Actually, I was informed this morning that we got IRB for Namibia, so I could change Namibia to green, to darker green. But we're also about to start, hopefully, in South Africa and in Zimbabwe, and also in Uganda. So we've currently recruited over 500 individuals, collecting blood and tissue, as well as detailed clinical information, and the sequencing is ongoing. And the work I'm going to show you next is... I don't have to go-- The work I'll be showing you, the analysis is done by those four post-docs from Yamachi. So very quickly, the demographics of our TACA participants are as follows. Majority of them so far are breast cancer patients, although we are now actively recruiting colorectal cancer. So the colorectal numbers started-- We started recruiting colorectal this year, so those numbers should be going up over time. And a small number of lung. Lung is challenging in our context. Not many biopsies are done. It's often confused with tuberculosis. But we figure it's worth looking. So we are trying to get some lung. But most of the samples we have so far are breast cancer patients. You can see, as you would expect, many people present late, so stage III and stage IV predominate in every country. And that is well understood in the African context, where people don't often present very early. Women typically show up with very significant tumors, unlike what you'd find in countries like the U.S., where there's a bit more early prevention. And you can see the average ages among countries, somewhere in the late 40s to early 50s. Although we have had, I think our youngest participant was 22 with stage IV breast cancer, which is also not that unusual in the African context. We are looking at subtypes, and here an interesting picture is forming. We all know about triple-negative breast cancer being predominant in this country in African American women. We're seeing a pattern where triple-negative breast cancer seems to be more prevalent in Ghana and Nigeria, two West African countries, as compared to Kenya, which is East Africa, which has more luminal B. You can see the Seychelles has hardly any triple-negative breast cancer. Obviously, these numbers are still relatively small. But once again, there's not that much data out there, so this also represents some of the first cancer epidemiological data of this kind, of this quality. So as we build up TACA to 1,000, 2,000 breast cancer patients, we'll have a very clear idea of the geographic distribution of subtypes, which I think might be a very interesting area of study. And so sequencing has begun. We're doing whole genome sequencing on NovaSeq X Plus. We're doing all of this in Accra. So we figured out the logistics and the regulatory sort of impediments in moving samples around. So all our samples come to Accra and are sequenced at our lab in Accra. And when you do stuff from Africa, you often have to answer the questions of quality and how good things are. So we just take that on head-on. I figured I'd put a couple of slides for that.Fine, we understand these questions are valid. So we're doing germline sequencing. Our target is 30X. You can see our median coverage is actually 41X. And for our somatic, we're targeting 100X, and we're getting a median of around 121. And you can see the tumor purity and homozygosity is pretty good. We are assessing tumor cellularity before we move on with sequencing, and so we're trying to skew towards high enough tumor cellularity. This is my most favorite slide in this deck. We decided to overlay our data with 1000 Genome and human HDGP data, really to kind of see is TACA data distinguishable from the "gold standard" data that's currently available. And as you can see here, it is indistinguishable. So the triangles are TACA data. We've overlaid it with data from these well-established databases. And as you can see, our data perfectly aligns with the African ancestry data that's already in these databases. You can see the Ghana and Nigeria samples cluster slightly closer together with the Kenyan samples slightly down. And then you can see the Seychelles data demonstrating admixture with South Asian, which makes a lot of sense. So basically, this is the sort of sanity check slide to show that sequencing done by Yamachi in TACA is on par with the data that already exists. And so at least at this level, I think we can be quite confident on how we move forward. We've assessed tumor mutational burden. There's a lot of conflicting literature about whether African individuals have higher TMB or lower TMB. We show that in our cohort so far in breast cancer, for the TACA cohort, we're somewhere just higher than TCGA breast cancer cohort, but still within a similar order of magnitude. Which would suggest that they would be candidates for checkpoint blockade therapy. Which I have seen papers talking about how you have to lower thresholds for Black patients because they typically don't have as much TMB. In this cohort, at least, our TMB seems in line with what has been seen. And then looking at sort of the mutational profiles. These slides are very fresh, so we're still analyzing data as more sequencing comes in. But we're seeing what appear to be differences in signature between subtypes, which would agree with what others have reported. And of course, as I said, the geographic distribution of these subtypes suggests that there may be some inherent differences from ancestry that determine which subtypes present in different places. We're also finding a number of enriched drivers. So our main interest in this is can we identify novel drivers in these populations that, quite frankly, have never been studied before? Some of these populations have never been sequenced. And certainly have never had cancer, breast cancer, or any other cancer sequenced. So this is also very preliminary, but we are showing TP53, so it fits with the idea that African ancestry individuals tend to have enrichment in TP53 mutations. But we are identifying a number of other potential drivers that appear to be much more enriched in our populations than they are, for instance, on the cBioPortal, which represents existing databases. So a lot more work needs to be done, but we are finding that there are certain targets that... or certain drivers that appear to be over-represented in our populations and underrepresented in others. And this is just a couple of them, where actually we're finding subtype-specific differences. So for LEP R or leptin receptor, we find that they're over-represented in TNBCs, but not in the luminal. Whereas for KRTAP10-6, it's higher in luminal versus TNBC. So this is all very preliminary, but I think we're showing that there are subtype-specific differences within our population, but there's also enrichment within our population relative to primarily European ancestry groups. So this is where I will conclude. I think the point I'm trying to make is including African genomic data in cancer research is not a matter of inclusion or diversity or whatever word you may choose to use. It is good science. Because there's a lot we don't understand about cancer among everyone, and understanding it better in these diverse populations will help improve how we do research in cancer. So whether it's identifying novel mutations for risk prediction, novel clinically actionable variants, better annotation of VUSs. There's a huge problem in underrepresented minority patients who are all classified as VUS. In fact, the reason the Ministry of Health in the Seychelles signed on to TACA is because they are actually a high-income country. They have universal healthcare. Everybody with cancer gets sequenced or gets a panel done in the Seychelles, and yet all the results they get back are VUSs. So they're really frustrated by that. So they agreed to sign on to TACA because they wanted to sequence these people to understand, "Why do we keep getting these? What are these variants that we're getting?" So that, once again, shows access. If you don't include it in the original data, even if you can afford these tests, they're of limited value because they give you data that's really not actionable. So I'll end here with talking about localizing research. It's important that we study underrepresented populations in this country and in Europe and other high-income countries. But we also need to work towards building capacity in these locations so that some of this research can be done locally. Because if all we do is study them in our labs in high-income countries, we will understand these diseases better, but we'll not build the infrastructure required to ensure that access to these end products of this research is available locally. And so that's work that we are trying to do at Yamachi and obviously with collaborators, including Eresh and CSIR. So I'd just like to acknowledge all the hospitals that have supported us across multiple countries. It's a big leap to work with an unproven startup. But they've taken that leap with us. We have a number of institutes that have also worked with us and also a number of... technology platforms and obviously Roche in partnering with us on TACA. But most importantly, the study participants, many of whom are late-stage patients with very few options, no access to clinical trials, probably a good number of our cohort will not survive very long from the time we see them. But they sign on and consent because they hope that this sacrifice will help someone else in the future, and so really want to honor that contribution. So thank you very much, and we'll hand over to Irechen for the next talk. Thank you. Hey, thanks, Yo. Thank you everyone for coming here quickly. So my talk is going to be a lot different from what Yo just presented, so let's get into it. No disclosures. I was actually going to stand here for two minutes and say absolutely nothing. Just let us take that two minutes and to realize that someone would've just lost their life to cancer on the African continent. Yeah, that's actually quite staggering. I think that's more than double what it is in the USA. By 2030, we expect that number to rise to 1.5 million new cases, which is also quite staggering. 70% of the global deaths occur in low to middle-income countries. We really carry the burden. You've heard this probably many times already this week. Less than 2% of the global cancer genomics databases come from people of African descent. That essentially means that entire continent is invisible to science from a genomics perspective. And here I am about to speak about proteomics, and if you think about that, there's probably, I don't know, 0.1% of all proteomics data sets represented globally. What we found is that there's around three to six times high mortality rates for some cancers versus the global average if you look at African patients. And this is mostly driven by late detection and biology that we really don't understand. And what we're trying to do from a proteomics level is to address that question. One, can we identify cancers earlier? And two, can we just understand what is really happening? What is different in our people versus the rest of the world? So you've heard this from Yo, and you've seen really nice data that African genetic architecture is completely different from the rest of the world. There's things that are very unique to the African continent, the environment, the diet, the microbiome, social determinants, and these are many things that we don't really consider all too well when we do science on the African continent, or when outsiders look at the African continent science being done. Proteomics is what I do and have been doing for the most part of my career. And this is something that can really bridge some of these gaps. It enables us to really get an understanding of the post-translational biology. What is really happening? It is a functional layer that bridges genotype and phenotype, and again, as I mentioned in the very beginning, the number of data sets for proteomics specifically on the African continent is almost zero. It's negligible. So what we found that works really nicely is that collaboration, particularly in South Africa, is great because there's really nice pipelines that have been built and biobanks, and we have actually the expertise to be able to address this critical gap. So let me just do a little bit of punting of proteomics. Why proteomics? So everyone knows why the genome is important. We have around 20,000 genes, where we can get some kind of information on variants and mutations, and the transcriptome, which is around 100,000 transcripts, where you can get an idea of the gene expression level. And then the wonderful proteome, which has something like a million proteo forms. And we have decided to work in that area specifically. So what is the advantages of proteomics and understanding the proteome? This really captures PTMs, post-translational modifications. It reveals active protein-protein interaction networks in real tumors. So this is not predicted stuff. This is actually measuring the effect of molecules. It gives you a direct readout of signaling states. And increasingly what we are finding in the studies that we do is that there are Africa-specific or Africa-unique biomarkers which could potentially lead to novel drug targets for African patients. So here I just want to touch on maybe four studies that have been done very recently. The common thing from all of these studies is that they've all been done through our research center at the CSIR, where we've been doing proteomics for the best part of 15 to 20 years now. So it'll be a gallbladder study, a prostate cancer study, pancreatic cancer, and breast. And really some of these cancers are some of the cancer types that are the most prevalent in Southern Africa and really have the most devastation. So, the breast cancer study. If you look at the five-year survival rate for the breast cancers, we have around 40% versus 90% in the US, which is absolutely staggering. Yo spoke about the subtypes that are apparent, but in our study, we found that triple-negative was the highest, and then HER2 was 22%, and Luminal A and B had 38%. And there were some distinct proteome changes in those types of subgroups. In the study, there was 28 tumor biopsies that wereprotein sequenced. And across the entire cohort, we found over 2,000 proteins. And what we really saw was that there was this massive increase in spliceosome activity and MYC targets. So that proliferation was just being really, really highly overactive. What we found was pretty cool and unique... See if there's a mouse. Yes. Was CD36 expression, which we found to have a fivefold decrease in abundance, which is quite different to what we see in the American populations. What was really cool about the proteomics data specifically was when we tried to do some machine learning based classifications, we had an AUC of 96% when we tried to classify tumor versus normal tissue. This is quite cool though, even though it was a really small study. And when we looked at the principal component analysis, again, this is all just on protein-level data, you can see that the proteome is really well-equipped to tease apart the different tumor versus normal patients. We then looked at a study on pancreatic cancer, where we have, again, a really poor survival rate of around 10%. And this is really largely driven by the fact that patients come through once they're at a really advanced stage, after the cancer has metastasized. Here, one of the things that was really important in what we saw was ECM remodeling, and there was a lot of proteins that were shown to be very unique to this cohort. And some of them were seen for the very first time to be linked to pancreatic cancer. We then got a bit lucky because that study was published in 2020, and we were able to get more samples from similar patients. But this time we looked at a plasma proteome. So in this study, there was increase in ECM remodeling again, which is a common theme that we see. And there was increased inflammatory mediators. And again, some of the proteins that we saw in this, and this now is soluble proteome, which is plasma proteomics, were again shown to be for the first time in African patients. Something that we really liked was the LGR1 AUCs were really high, and it showed really nice correlation with... Excuse me. Yeah, so there was AUC of 0.94 for LGR1 when we were comparing pancreatic cancers versus benign pathologies. These are all really nice datasets, but they are kind of small, right? So the plan is now to really go into much bigger datasets. So another study that came out from our lab was a prostate cancer study. Here, unfortunately, African men carry a really high incidence and aggressive phenotypes. We predict there to be around 4,000 new cases per annum. And in Southern Africa, for some reason, there's the greatest incidence across the entire African continent, where 65.1 men per 100,000 will be diagnosed with prostate cancer. So what did we see in this study? This was done very recently in the last two years. Here we analyzed over 150 patient clinical samples. So these were actually tissues from patients. We identified around 3,200 proteins in total, and then of those, 130 were significantly different between cancerous and benign patients. 12 of these proteins were linked to nerve function, and this is something that the collaborator was really interested in. She wanted to know what was the involvement of the nervous system in how cancers metastasize. And what she was really excited about was that there was a five-protein panel that kind of confirmed her initial hypothesis that cancer is being driven by this neurogenous metastasis process. And then lastly, we looked at a gallbladder cancer in 2023. Here as well, the five-year survival rate is really poor. So there's a lot of common threads that are being shown in these studies. This study was done in collaboration with Wits University. And here we found that there was nine key proteins linked to gallbladder cancer, and there was, again, extracellular matrix issues, and smooth muscle contraction loss. So I guess across all of these studies, there's quite an easy pattern to be able to extract, right? ECM remodeling is a central driver. There's dense tumor stroma, and it's consistent across a lot of the cohorts that I showed. There's a few proteins that are quite unique to this population. And the question really is, is ECM stiffness really driving chemoresistance and metastasis? And I guess further to that is because there's this kind of shield around the tumor, are we then unable to get drugs into the tumor to actually exert the effects? The second one is that there's a dysregulated lipid and retinoid metabolism. There's altered metabolic tumor environment. And here, as I mentioned at the start, this could be due to some kind of dietary obesity and diabetes, which may play a role in this specifically in the South African context. But the value here is, is there a metabolic vulnerability that allows us to repurpose some of the drugs that already exist?And I guess finally, and this is also now an extension of what Yaw mentioned, but this is from the proteomic side, is that there's population specific proteomic heterogeneity. There's African specific patterns that are really absent from Asian and European cohorts. And if we extrapolate global studies to the African patients, it just really doesn't make any sense and doesn't work. What we're finding increasingly is that looking at single protein markers doesn't really help, and looking at multi-analyte panels is the way forward. So Africa really needs its own reference data sets. So let me introduce you to what our answer is to this. So I hopefully have shown you that proteomics is useful, although in very small studies, and has illuminated some cancer specific biology, particularly in Southern Africa. But what we know for a fact is that tumors are very complex, and no single omic technology can get a full picture of what is really happening. So I'd like to introduce you to the African ClinMap, which is the African Clinical Multi-Omic Atlas Platform, which is a program that is run by the CSIR. It's only been probably a year and a half since its inception, so still fairly new. So let me just set the scene for what the African ClinMap wants to do by using a little story. So if we wanted to drive from Johannesburg to Cape Town all the way down here, historically, we would have to pick up a map and plot our way and just get in the car and drive. We wouldn't know where the stops were, where the detours were, where the potholes were, none of that, right? But lo and behold, we have Google Maps now, right? Or Waze. So thanks to this, we can basically plot our entire path, and we will know where the stops are, where the traffic cops are, where there's a detour in the road, where to stop for food, and so on. So it's against this backdrop that I think I want you to just keep this in mind, and I'll come back to it to kind of tie in how the African ClinMap wants to work. So it's really a strategic multi-omics research program. I've spoken a little bit about proteomics and what's so great about it, in my opinion. Yaw has spoken a lot about genomics and transcriptomics. But essentially, what we want to do in this platform is to tie these three disciplines together. What that allows us then to do is have a lot more data to enable us to make more clinically informed decisions. So the beauty of this platform is that all of these technologies-- or we have expertise in all of these technologies housed at the CSIR. So we have, I guess, experts in transcriptomics, single cell profiles, genomic sequencing, and I've shown you a lot of the clinical proteomics work that we've done. Now, our idea is to get clinically defined African cohorts and really get multiple layers of information from them included with their clinical outcomes, and integrate this using AI and machine learning. So what this will enable us to do is have an African relevant digital health resource, which hopefully will inform novel diagnostic and drug targets, and what this ultimately boils down to is a data driven health innovation. So to come back to this picture of driving or our map again, having all of the access to the data will enable us to do one main thing. So if we're on this green line, which is our path of health, and we're happily moving along without any issues, that's the best case scenario for everyone, right? But we all know somewhere along our path of life, we can end up moving towards some kind of disease state, and in this instance, we're looking at cancer specifically. Our hope is that by having access to all of the different data layers, we can have clinicians have much more data available to get us back to that path of wellness. Now, the ideal situation would be to really get us early detection, or really back onto the state of health at a very early stage in our disease lifetime. So, as Yaw mentioned, Africa is the most diverse place on Earth. There's over 3,000 distinct ethnic groups on the African continent, and we are home to the most genetics that anyone has ever seen. So the ClinMap platform wants to address this by really doing deep phenotyping, longitudinal sampling, multiethnic cohort recruitment. So, so far we've mostly done our recruitment in South Africa, where we have the Bantu, Khoisan, Cape Malay, people of Indian descent as well included in some of the cohorts. And we will have computational deconvolution powered by AI. We've partnered with a company called Prompt Bio in San Francisco, who are really a pioneering AI based platform that is built from the ground up specifically for multi-omics data integration. And we will run the entire platform through an open data based on the FAIR principles. So Yaw, again, mentioned precision medicine, and this is the vision for how we intend on doing this through the platform. Patient biobanks exist. There's a lot of samples existing on the continent. Sometimes they might not be the best fit for what we need to do, so we might have to do recruitment. However, we have the in-house technologies to do matched blood, urine, and tissue analysis on the transcriptome, genome, and proteome levels.And then using the platforms that I just mentioned, do data integration, look at understanding the biological pathways and subtypes of these different cancers that are unique to the African population, and ultimately have some kind of clinical translation where this actually will benefit a patient. Because it's all good and well, and typically happens on the African continent where people do these studies, but it doesn't really go anywhere. So the ClinMap target is to have 5,000 to 7,500 cancer patient multi-ome data sets in the next two years. And if you translate this to data points, it really explodes because on each of the different omics technologies, you have different levels of data points. But we're looking at over 100,000 data points per patient in this data set. So to achieve this, what we've been able to do so far in this one year is set up a clinical network across Southern Africa, in Namibia, Zimbabwe, and South Africa, and very recently partnered with Yiamachi Biotech in Ghana. And like I mentioned, our AI integration partner is Prompt Bio. So hopefully, we'll be able to spread across the African continent and do this multi-omic profiling in more and more patients. But I think the first port of call is for us to have set up this network and show the value of this on really small or small scale studies for now. So some of the first work that is currently ongoing as of last week is a cervical cancer cohort where we will look at this multi-ome profiling. This is in partnership with Stellenbosch University, and we have been doing some work, which has been funded by Genentech, on colorectal cancer in partnership with Wits University. So I guess cancer biology needs to be understood. We cannot just let Africa be a gap, and this is a message that we've got through quite strongly already from Yao. What we hope to get is diagnostic innovation, so Africa-specific biomarkers with deployable point-of-care tools and some kind of cancer risk scores when we combine these multiple layers of information. We hope that this will inform therapeutic strategy, where we can show that there are novel targets for Africans based on their tumor protein profiles, repurposing of drugs for local variants, and at the very end of the program, we hope to have personalized medicine roadmap for African patients. Yao touched on this as well. We would love to build African omics and bioinformatics expertise through all of this data that we will generate. We've shown a Southern African collaboration model and hope to extend this beyond Southern Africa into the rest of Africa, and we purposefully built this with data sovereignty and community benefit as one of the pillars of the platform. I'd just like to conclude with these four. African biology is unique. We have unique discoveries that need to be made, and I hope I've shown you some of that with the proteomics data, and I have no doubt that by integrating all the other layers of information, that this will be even more true. The ClinMap platform is infrastructure built. We've spent many years building the infrastructure around the different omic technologies, and this has allowed us to now gain some traction into how this will all work together. We know that multi-omics is the next frontier. We've heard this many times so far this week, and we really hope that we will have the opportunity to integrate the proteomics, genomics, and transcriptome data to fully capture the biological heterogeneity that's unique in African populations. And ultimately, and I think this is most important for all science that's done, whether it's on the African continent or not, science needs to be done to serve patients. Here we have built our platform, and it's designed in a way that really has the patient at heart. Whilst they might not see the benefit immediately, it is really imperative that they understand that the platform exists to benefit them. So lastly, African biology is unique. I think we're getting this message quite clearly across, and it deserves world-class science. I just want to acknowledge all of our partner institutes that have contributed to the platform. This is my other PIs, and you can see our lovely Maspec in the back here. There's an African proverb that says, "If you want to go fast, go alone. If you want to go far, go together." And really here, we're showing the power of people working together across the African continent to better enable health innovation for the African continent with far-reaching implications, and as Yao mentioned, everyone emanated out of Africa. So ideally, if we can understand most of the biology on the African continent, it more readily can be extrapolated to the rest of the world. If you want to find out more about ClinMap and what we do or how to get involved, you can scan this 2D barcode, and it'll take you to our website. I just want to thank our funders, some of them for infrastructure, some of them for actual grants, our Department of Science and Innovation. And I just want to thank AACR as well for affording me the opportunity to present this. And, yeah. Thank you to Yao for the invitation. And yeah, I'm happy to have discussions afterwards. Thank you. Wow, I really should have begged to go first. They've really set a high bar here. Okay. Good morning. We're almost at the end of the conference, so thank you once again for attending and for your attention, and of course, to the chair of the session, Yao, for the invitation to speak. So my name is Nyasha Chambwe, and I'm based in New York City here in the US. But I think as part of this session, we're all interested fundamentally in the same problem, which is the fact that Black people tend to get the most aggressive forms of different cancer types and have the worst outcome. So I'm going to take a slightly different track this morning and talk about one type of cancer and go really deep into our efforts of understanding how ancestry-related biology impacts endometrial cancer disparities. These are my financial disclosures for the conference. So endometrial cancer, these are cancers of the inner lining of the womb. Despite progress in many different cancer types, endometrial cancer has the claim to fame that it's actually increasing in incidence and mortality over time, despite all our best efforts. As represented in this data from the US, the surveillance database, we see also that endometrial cancer has stark racial disparities. As you can see overall, across all racial and ethnic groups, incidence is increasing over time, but we're observing that the sharpest increases are happening in the non-Hispanic Black population. Similarly, when we look at this mortality data, it should be extremely striking to you that the mortality gap between non-Hispanic Black women and all other racial and ethnic groups is quite stark, more than two or threefold, and is continuing to broaden. We're really interested in understanding the underlying biology driving this disparity. And so there's a lot of heterogeneity within endometrial cancer. It's historically classified according to histology, with the predominant histological type being endometrioid, and then non-endometrioid subtypes of serous, clear cell, and carcinosarcoma, which is not represented on this slide with all these characteristics. The non-endometrioid subtypes, which are also primarily high grade, are the most aggressive subtypes. In the age of genomic sequencing and through TCGA, endometrial cancer now has four established molecular subtypes that are defined by poly mutation, microsatellite instability, and then having high or low copy number variability within the subtype. What's striking is that these subtypes also define the mortality trajectory of that tumor, and once again, we see that these non-endometrioid histological subtypes and the copy number high subtype, which is the most aggressive and has the worst mortality, is unsurprisingly linked to Black women. So Black women tend to get non-endometrioid subtypes or the copy number high subtype, which is dominated by TP53 mutation. I think this ties in nicely with the data that Yao showed from Seethaler's recent meta-analysis that showed an increasing prevalence of T53 mutation in an African ancestry background. As is the theme of the session, our molecular understanding of endometrial cancer is actually built on cohorts that underrepresent the population that is the most affected, either in incidence or mortality. So I'm not picking on any particular cohorts, but these are some of the three largest published cohorts in endometrial cancer: The Cancer Genome Atlas, the NIH initiative, the CPTAC, which is genomics and proteomics as well, as well as a recently published clinical panel seek cohort of around 1,800 patients from MSK. So you can see here these cohorts represented in terms of racial distribution, and it becomes readily apparent that they are primarily cohorts that are dominated by White women, and also that these cohorts are dominated by endometrioid histological subtype at recruitment. So we really don't understand the underlying biology of the most aggressive subtype in the population most affected by those aggressive subtype. So what I'm going to present this morning is an initiative we started in New York, where we are based, as a collaboration across many institutions in the New York region, taking advantage of the diversity and immigrant population of the New York City metro area. So represented here on the map, you can see that we have enclaves across New York City of foreign-born individuals of really diverse racial and ethnic subgroups. So this initiative led through the New York Genome Center looked to enrich and recruit participants across different cancer types to try to understand underlying biology of cancer disparities.And so I'm going to present to you today work from that cohort in endometrial cancer and juxtapose against those public cohorts I just showed you. We see that we made an effort to enrich our recruitment for Black women who self-identify as Black and non-Hispanic, as well as trying to diversify the histological subtype in the cohorts to more mirror the more aggressive subtypes, non-endometroid subtypes. Just as a segue, I'm going to focus today on the biological determinants, but suffice it to say, there's not enough time today to discuss the complexity behind all the multifactorial factors that influence cancer disparities. And there's this push and pull that we often experience because we know some of these environmental and psychosocial factors are usually more or less correlated with how people self-identify by race or ethnicity, which influences some of their lived experience. But on the other hand, we think a lot about how tumor biology, as a biological construct, perhaps more readily lends itself to comparisons based on genetic ancestry that is inferred directly from people's underlying genome. So I'm going to more or less talk about the right side today in the 20 minutes I've been allocated, and in the study that we have ready to go for publication. So in this phase one of the study, we have 71 patients that were sequenced, mean age at diagnosis 68, and again, a predominance of self-identified Black or African American women who are non-Hispanic and with an enrichment across non-endometroid subtypes. So we were able to procure normal and matched tumor, partial normal tissue, and PBMCs for a subset, and then these samples underwent whole genome sequencing as well as tumor-only RNA sequencing, and then we performed some downstream analysis from there. I'm going to describe some of our key findings from this study. Initially, we used a germline whole genome sequencing to infer global genetic ancestry. As you can see here represented in yellow, the vast majority of the cohorts, in line with their own self-identification as Black, was of predominant African ancestry, which is relatively unheard of in any published endometrial cancer cohorts. We also looked at the subtype subcontinental level. So not just the continental, but subcontinental. Herein, we focused on the subcontinental African ancestry subgroups, and we saw that relative to TCGA, we had a paucity of Yoruban ancestry in our cohort, which is quite striking and represents the heterogeneity in African ancestry populations when we look at them at a high level. You can just assume, "Oh, this is the same. They're likely the same." TCGA cohort was also sampled within the US, but here we saw that even though all of these populations were sampled within the US, the underlying subcontinental diversity in terms of African ancestry was really quite different. Similarly, making a case for looking at ancestry versus race for biological determinants analysis, you can see that sometimes you have a discordance from what you would expect between how somebody self-identifies by race relative to their inferred genetic ancestry. Okay. So, for this presentation, I'm going to switch it up a little bit from how I normally present it and show you guys just a brief outline of the findings we had from the RNA sequencing data. So here again, this is our attempt to leverage multi-omic data to understand disparities. So we also have an unpublished normal endometrium single-cell reference from diverse donors, and we leverage the bulk RNA-seq from the cancer genome data to deconvolute, to try to understand what cellular lineages are in our cohort. And in line with what we would expect, we had a high prevalence of epithelial cell types across the different molecular subtypes. One of the subtypes we have here, the carcinosarcoma subtype, is known to have a mesenchymal, somewhat sarcomotous component, which includes epithelial and mesenchymal. And we are able to show through single-cell-based deconvolution that indeed the E to M or epithelial-mesenchymal ratio was much lower in that subtype. We performed differential expression analysis relative to ancestry. We found three really robust differentially expressed genes relative to the amount of African ancestry. So here we use the numerical admixture coefficient on the x-axis. And you can see that for two of these genes, the higher African ancestry component you had, the lower the expression, and then the opposite for PLXNA3. When we expanded this to look at the pathway level, we found interesting, enriched, distinct gene expression programs based on ancestry within endometrial cancer, even after controlling for histological and molecular subtypes. Again, building on the story we've been hearing this morning, that there is an ancestry-specific biology that could perhaps inform biomarker discovery andtherapeutic target selection. There is literature and precedent about looking at how genetic ancestry shapes the immune response. There's extensive data from infectious disease that shows that there's an enhanced inflammatory response to pathogen exposure with increasing African ancestry. And similarly, the study that Eyal referenced which showed that the total lymphocyte infiltrate levels differ by ancestry in different cancer types. In this case, the precedent was set in breast cancer but hasn't been fully established in endometrial cancer. So we leveraged our data to perform deconvolution primarily because immune checkpoint therapy has become standard of care for high-grade, more aggressive endometrial cancer and there's data to suggest that there are specific immune-based subclusters within endometrial cancer that influence outcomes. And so we used our RNA-seq to deconvolute and estimate the relative prevalence of different immune cell types in this cohort. And suffice it to say, when we took those estimates and built linear models that correct for tumor purity, molecular and histological subtype, again, accounting for the underlying endometrial cancer heterogeneity, we found an interesting relationship with CD8 T cell infiltration was much lower, depending on your level of African ancestry or the coefficient of African admixture. And here we're using the coefficient specifically because we are looking at an admixed African population where the relative amount of potential proportion of ancestry attributed to being African is different and so this tends to work better in analytical models. We're able to show that here, that higher African ancestry individuals tended to have lower CD8 T cell infiltration, which we also saw with a related TGF-beta signaling pathway that was downregulated in African ancestry. We tried to validate this finding orthogonally in the TCGA cohort by building similar models with TCGA deconvolution estimates, but we were not able to fully recapitulate the global CD8 T cell finding and we think this is primarily because those cohorts are in and themselves fundamentally different and our cohort is more enriched for high-grade non-endometroid tumors. But we still were able to find suggestion that naive CD8 T cells are decreased in prevalence in African ancestry endometrial tumors in line with pathway analysis that support immunoregulatory interactions being enriched with African ancestry. So those were primarily findings from the transcriptomic integration. So I want to turn our attention back to the whole genome sequencing that we performed. So we found after classifying the tumors by their molecular subtype, the majority of the cases in our cohort were of this copy number high subtype and the main findings we found in terms of genomic alterations are here. The genes highlighted and starred are either under selection or have been previously reported oncogenes or tumor suppressor genes that are drivers in endometrial cancer. Interestingly enough, from the TCGA work showing higher prevalence of FBXW7 mutations with African ancestry in pan-cancer, we actually found that back here in this cohort again where we saw enrichment of FBXW7 mutations, particularly in the copy number high subtype. Because our cohort was primarily of African ancestry, we analyzed it jointly with the TCGA to try to understand differences in driver mutation frequencies between ancestries. So here again, controlling for histology and molecular subtype, we identified genes that were more enriched in mutations in African ancestry. Some of them have been previously reported such as FAT1, SETD1B in the endometrioid subtype and then SPOP in the serous subtype which is also an interesting ancestry-associated driver in prostate cancer, EE2F in carcinosarcoma and copy number high tumors as well as also genes that were potentially depleted in African ancestry. So what do all these comparisons point to? They point to the fact that many of these known drivers are either current targets of treatment or are used to stratify individuals for treatment and so these kinds of analyses have important implications on the question we are looking at today, how we are able to implement precision medicine for all. Again, in the similar vein, we looked at the OncoKB database to assess the mutations in our cohort for clinical actionability. And what we see, which is not surprising at all given all the data that came before to populate the database, we have fewer clinically actionable mutations in the African ancestry individuals in our cohort relative to European ancestry and we did this by combining our cohort with TCGAwhich was where we see the statistical significance. I think this is a power issue, but we don't see it when we subset it out, although we do see it within TCGA itself. Beyond the single nucleotide variation, we also had the luxury of having deep coverage whole genome sequencing which allowed us to go very deeply into the structural variant landscape of these tumors. So off the bat, here is the copy number high subset. You can see that there's an extensive structural variation, altered structural variation landscape across these tumors. The copy number low subtype here in blue is relatively quiet, and we do see similar genomic instability across the microsatellite high and the poly mutant tumors, which are very well known to have good outcomes and high tumor mutation burden, particularly with immunotherapy. And so we went deeper and looked at focal amplifications and gains across the whole genome. And so this is the entire landscape. I won't have time to go through it, but I want you to zoom in on MECOM, which was this locus on chromosome 3q, which was the most frequently amplified and deleted region. So MECOM, MDSC1, and EVI1 complex locus, it's a great name for a gene, is a master transcription factor that encodes two protein products. So it has a long isoform, which has been shown to have tumor suppressive features. But the short isoform, EVI1, is oncogene implicated, especially in leukemia and ovarian cancer. So you can see these two isoform variants here on the chromosome three location in this about 1.5 megabase region. When we overlaid our data, including by ancestry and molecular subtype, what we were able to show, which is not really well shown here, but if you look here to the right, we saw that the focal amplification events tended to be concentrated over the short isoform relative to the long isoform. And the focal amplification was driven by diverse mechanisms of structural variation, including examples represented here, whether it was duplication, parago, or ecDNA or dimorphic mutants but also that this was more prevalent with African ancestry. So we looked at expression globally and saw that there's a slight increase here in expression if you have the focal amplification, although it was not statistically significant. Again, this might be simply a power issue because of the size of our cohort, but across both our cohort and the TCGA cohort, we saw that the prevalence of this MECOM focal gain was much higher at around 35% to 36% in African ancestry, compared to around 10%, 11% with European ancestry. This difference was, of course, statistically significant. This brings up a key point here where we assessed-- We don't have outcome data yet for our cohort, but when we looked at outcome data relative to MECOM status in the TCGA database, we were able to show across the board that for overall progression-free and disease-free survival, the MECOM amplification status was significantly associated with clinical outcomes. However, when the difference between the large gain and focal gain didn't seem to be statistically significant here, although our numbers, our N again is quite low. But this definitely has very interesting implications and is a potential novel biomarker or target in endometrial cancer. So to summarize our findings, I hope I've convinced you that in a deeper level that ancestry can shape the molecular landscape of cancers and in particular here, high-grade endometrial cancer through multiple biological axes. We recruited a cohort that was enriched for women with high-grade tumors who self-identify as Black, who are predominantly of African ancestry, and performed one of the largest studies with paired whole genome sequencing and RNA-seq in this setting to try to understand what are really the drivers of these more aggressive tumors that are more prevalent in Black women, both in the US and it's also true across the African continent. We have three main findings. The decreased CD8 T cell infiltration associated with African ancestry, the fact that African ancestry patients harbored fewer clinically actionable mutations, and this potentially exciting finding of the MECOM short isoform focal gain being more associated with African ancestry, having an association with worse clinical out-- and having an association with worse clinical outcomes. There's still more work to do, of course, that we need to functionally validate these findings as well as dig deeper into our tumor microenvironment, CD8 T cell infiltration finding. But more globally, of course, we need to expand these cohorts. I think we recruited over 100 or so patients, but of course, by the time you are reporting findings, you lose some patients along the way due to quality control and other such issues. But just in this setting, we were able to show applicable novel biology of potential interest, hinting to the fact that if we want to able to really understand the underlying biology of disparity, we need to be really intentional in our recruitment of the populations most affected by the tumor types.So I'd really like to thank all my colleagues. This is not a work that you can do on your own. Our lead clinician from my hospital, Northwell Health, is Dr. Marina Frima, who's a gyn oncology surgeon. We have a vast network of collaborators, including Samir Beyaz, who is the wet lab lead on the project, as well as a strong collaboration with the New York Genome Center computational biology group, and we all work together jointly on this project. And I, of course, have wider acknowledgments to my group, our network of collaborators, and all our funders for their support. Thank you so much for your attention. Thank you very much, Nyasha. So I think now we have about 10 minutes left to open up for questions. So if you would like to ask a question for the benefit of our online audience, if you could use the mics. So, I'll throw it out to the audience, and I think online, they have the option of typing questions into a chat. Right now, there are none in the chat, but I'll keep an eye on that. So are there any questions for any of us? Thank you. I guess it's a race to the mic. I won. Hi, Christiana Friedenberg, University of Florida. I have a question for you, Dr. Chamwe. Did you actually know the percent of ancestry that yielded that CD8 positive or low expression or high expression CD8? So is it like 20% African ancestry you'd see it, or 70%? That's my question. Thanks. Okay. Thank you. No, not exactly. It's an overall... Because we had to control for molecular and histological subtype, it's difficult to give a threshold for what level of African ancestry merits the decrease. It's just across this full spectrum of the African ancestry coefficient from zero to 100%, overall, the CD8 T cell infiltration rate is lower. Understood. Thank you. Robin Lorenz, Genentech Roche Pathology. Obviously, incorporating all of the transcriptomics, the genomics, the proteomics is clearly what we believe in at Genentech and Roche. But have you also thought about the fact that there's a lot of data in your spatial components of your tissue, and either looking at the H&Es for foundational models or spatial transcriptomics, spatial proteomics for incorporating that data? Yeah. So that's a very interesting point. Just walking around the conference here, when people talk about proteomics- Your proteomics is very different- Yeah ... to my proteomics. I went to a lot of posters and then talks, and I was like, "Okay, this is completely different." So we haven't really considered that, but I think that's a component that we want to add because there's a lot of, as an example, FFP slices for many of these tissues that we want to analyze. And routinely, clinicians would have that analyzed by some kind of pathologist. So our idea is to also embed another layer of data into the multiple data types that we will have, and one will be image based. Okay. And, yeah. We're, again, not exactly sure how that will tie in with all the other omes, but there is a plan definitely to include that. I think it's very valuable. Mm-hmm. Sure. Yeah. And with TACA, the vision is similar. In fact, why we're looking to collaborate with CSR is to add additional omic layers. And spatial is one that we're very keen to pursue. We're trying to figure out how we can do some fresh tissue, as well as FFPE, to expand what we can do with the tissue. Obviously, in our context, fresh tissue is sometimes a luxury. But these days, there are new ways without needing liquid nitrogen and the other kind of preservatives that we could use. So we are exploring that and looking to see how we can integrate that. And actually, at the level of H&E, we're already starting a process where we're scanning all the slides that are coming in before we do any omics. So we will have at least H&E stained images that could be the basis for more in-depth spatial work down the line. So that is certainly something, actually on the prompting of your colleagues at Roche Genentech, so yeah. Yeah, because I think nowadays with some of the new generative AI and stuff, you may be able to do things with that H&E image that- Okay ... you don't need the more expensive spatial to do. Okay. Hi. First of all, thank you so much for your talk. I was wondering, a lot of the multi-omics data sets that you mentioned, such as TACA and ClinMap, you mentioned that they're very useful for personalized cancer medicines, but especially with the high cost associated with those, and they may be inaccessible to certain populations. Do you believe there's an applicability in broad population applicable therapies, and how multi-omics data sets can be applicable to those? Yeah. I can start and let Irusha and Onyasha weigh in as well. I think you're exactly right. It's all nice to have a really tailored... I've been in some of the talks today or this week, and it's been amazing what can be done with a vaccine that is tailored to one individual. But the reality is that's barely cost effective here, probably not cost effective here, and certainly not cost effective in the context in which we work. So I think for us, we're always thinking of how do you identify broadly applicable solutions. And I think I would argue that's another way that the diversity of Africa is useful, because Africa is the one continent where you could subtype individuals or subdivide populations and look for common, whether it's antigens or other markers that would be unique to perhaps, maybe not the whole continent, but maybe a particular group. And so I think Africa provides the opportunity of doing that kind of science in a single location. Irusha is working in South Africa, one of the most multiracial countries in the world, where you could, in a single hospitalReally, a bit like New York, I guess. But you could really explore some of those. And so I think Africa as a continent provides that opportunity, and so I think it's important that... Actually, a good question you're asking us. As we are working, we don't necessarily get too drawn into simply chasing a single antigen for a single individual, but think about how we could really take a step back and say, can we actually push science to now find more general solutions that are more broadly applicable and ultimately would be more accessible to more people? So I think that's a really good point and one that we're certainly keeping in mind. Yeah. Yeah. So from our side, on the ClinMap perspective, there's no data, right? It doesn't matter which level. And we want to build that data that can firstly be very small. As I showed, there was a lot of data sets that we have that's very small, under 150 patients or so. Some of them even smaller than that, 14, 15. But it needs to have that, as you mentioned, generalizability. Yes, you might find things that are very unique to that subset, but what was also quite cool in what I showed was that across different types of cancers, there was a common thread of what's happening on the African continent. So I think those are things that can definitely be explored and show that generalizability that you're talking about. Thank you. Any other questions? I had one, Nyasha. Watching some of the talks yesterday about, it seems whenever people talk about African ancestry tumors these days, everything is immune infiltration and it's CD8s. I think it was prostate that we heard about, and now you're seeing this there as well. What implications do you think this has for immunotherapy? Does it mean African people could be a... Because obviously access to immunotherapy in Africa is very limited right now. Mm-hmm. But do you think these make the case that we should be looking to improve access to immunotherapy, that these populations could be exceptionally well served? Or do you think on the flip side, it means some of these may not work as well, and we need to think about other solutions? I don't know if you have a clear answer, but yeah. Yeah. I don't have a clear answer, especially when I have to answer the question from an immunologist. I will say that I think just the immune landscape across tumors in relation to ancestry has become very interesting because there's also a component of response to environmental exposures as well baked into it. So we've begun to really think about it. That's why I showed that slide about how African ancestry is related to pathogen exposure. There's diet and other exposure relationships with the changes in the immune microenvironment. At the same time, those loci are also controlled by the most genetically diverse genomic loci, like the HLA, for example, which we know historically one hasn't been mapped well. The HLA haplotypes on the African continent are probably not complete in our HLA databases, et cetera. So I think it is an area that is ripe with knowledge gaps that need to be addressed with more data from the relevant populations, but also the right data. So even though we can collect samples in and of themselves by... I think it's become really important to also keep track of the context in which those individuals lived in relation to how that influences the tumor microenvironment. So that's my non-answer. All right. Thank you. I think we're right at time. So, I think we'll draw it to a close. I'm sure we'll be available to talk a bit more. But thank you again for making the time to join us. I hope it's been a stimulating session and look forward to hopefully having additional sessions at AACR in future years where we can begin to show how these databases may begin to change how we think about cancer, not just across Africa, but really globally. So thank you very much for your time.