 Okay. As you probably can tell, I'm going to be talking about genetics. If any of you know you're Mark Twain, you may have figured out that I'm also going to be talking about statistics, so I hope you had your coffee with lunch. I do have a couple of conflicts. Not that either of these really make me any money, but I am the co-founder of two companies that are essentially grounded in the idea that you don't need to know your genetics in order to get dramatic improvements in your health. So I've wanted to do a talk like this for at least a couple of years, and it's been through various iterations. You're just going to get the one that happens to exist currently, and it started when I went to an event a couple of years ago, and I was listening to somebody who's big in the space that we're all in, considered an expert in genetics, and they said something along the lines of I'm a low carb genotype, but my partner is a low fat genotype, and I was like, hang on a second. Is that something we even know? This is like seriously, and I figured I would know that if that was a thing, and I didn't know that. However, I was willing to admit that maybe I didn't know about this enough, so that sort of led me on a bit of an odyssey to try and figure some of these things out. So this sort of starts with my why. Why would I even want to think about some of these things? And I think questions I would like to answer, and maybe help other people answer, is what does a change in gene function mean? What do our genetics mean? How do common genotypes that we talk about, say if you measure, you do your 23andMe, put it into some kind of tool, how do those genotypes alter phenotype? And phenotype is a thing that we can measure, say your blood glucose or your BMI. And then if you're going to make a recommendation, so I am a practitioner of sorts, and many people listening to this may be practitioners, do genetics change how we would recommend interventions based on genotype in addition to or separately from just measuring the phenotype that we're trying to change? So this kind of reminded me of this quote, and we're not really sure it came from. Mark Twain attributed it to Benjamin Disraeli who's a former British Prime Minister, which is that there are three kinds of lies, lies, damn lies, and statistics. And Wikipedia says that this is a phrase describing the persuasive power of numbers, particularly the use of statistics to bolster weak arguments. And this is something that I think we see a lot of in the genomics space. But I will warn you that I'm about to use statistics to bolster my arguments. So make of that what you will. However, when we think about genetics, the way we talk about it, we really are talking about statistics. So here are some of my own data, and I will be using my own data throughout the talk. If we think about methylene tetrahydrophodate reductase, MTHFR, my genotype, I'm told is associated with a 53% loss of function, and therefore I need twice as much codeine as somebody else every day. Or COMT, my genotype, apparently, is associated with a 38% increase in function. This means I have less dopamine in my dorsal lateral prefrontal cortex, which I'm sure you are all very concerned about. And what about genetic risk of diabetes? So I have a SNP in my melatonin receptor 1B, and that increases my risk of diabetes by 67%. I mean, that's quite a lot. Should I be worried about that? How much do these statements really apply to me? How much do they apply to you when you put your data through these tools? So a couple of notes. I'm mainly going to talk about the best research SNPs, single-nucleotide polymorphisms that you see in common genetic tools. There's not going to be any uncommon genes that have a big effect, and also no dominant or excessive mutations associated with significant disease like cystic fibrosis would be an obvious example. I've put my data into two online nutridonomics tools and then extracted some of the data, some of the papers. These are tools made by people who I generally respect and the tools that I imagine some of you have used. And the reason why I did that is because these are people who I think will probably understand it as well as we can understand it and then explain that to other people. However, all genes and statistical trickery are my own. And hopefully, there's a step-by-step process that you or your practitioner can use to think about risk. So just very briefly, for those who aren't familiar with it, we're talking about single-nucleotide polymorphisms here. You see all the nucleotides, adenine, guanine, thymine, cytosine. These are what make up your DNA. And if you have a change in one of these, you might get a change in gene function, and that may change your disease risk. Some statistical housekeeping before we start. It's important to know that most studies, genetic studies in particular, but most studies in general, make two assumptions about the data that they're working with. The first is that the data are normally distributed, so they follow a standard bell curve. And the second is that SNPs or genes that have similar effects are additive or linear. So if you have one copy of an allele, you can have up to two. If you have one copy, that causes a certain amount. If you have two, it's twice as bad. Or if you have multiple genes that affect a certain phenotype, they can all be added together. We do this because the data is much easier to analyze. But if you want to then try and figure out what's going on, it also makes it much easier for us to do the detective work to figure out what's going on. So that's nice. So this is just a standard normal distribution. Many of you will be familiar with. It's described by the mean, which is in the middle. That's the average. Most people are going to be around the mean. And then the standard deviation gives you the variability in that data. So if you go one standard deviation above and below the mean, it's about two thirds of people, a little bit more. And then as you go, two standard deviations, about 95% of people and so on. However, if we know the mean and the standard deviation, we can recreate the full bell curve, which allows us to see the full scope of what people might look like with a certain genotype. If you're looking at papers, and you're trying to interpret some of this stuff, then you'll see that sometimes people use the standard error of the mean instead of the standard deviation. And that's something that we can actually calculate, we can calculate the standard deviation if we know the standard error of the mean and the number of people. So just to give you an example of this, in the bottom left hand corner, I've taken data from 20,000 women in NHANES here in the US, and I've given their fasting blood glucose with the mean. And then the standard error of the mean is the error bars with increasing age. And you see, okay, so women who are 18 to 25, their fasting blood glucose is about 85. And then that increases until you're sort of 60 odd and older, and then it's sort of nearly 110. You think, okay, that's that's that's a really nice trend. You know, you clearly as you get older, your blood glucose goes up. What if I plot it with a standard deviation? It looks like this. So this, each of those error bars now cover 68% of people in those groups, and you still see the same trend in the middle dots, those the averages in each group, but it's much more variable. So when you're looking at papers and you're looking at graphs, the main reason why people use the standard error of the mean is because it makes their graphs look much nicer. However, that really hides a lot of the data hides a lot of the information that you need to interpret these studies. So the kind of graph that I'm going to use that I'm going to show you is called a violin plot. This is the same data you'll see there's a huge amount of overlap between all the age groups of women and fasting blood glucose. The full tails give you the full range of numbers that you could get. But you still see that the width still shows you the bell curve. So it still shows you the normal distribution. So when I've taken studies, I've tried to recreate the bell curves so that you can actually see the full the full scope of the data. And I've done that by taking the mean plus whatever error bars they used, and then generating a random data set. If I was putting my professor hat on, I'd say I've generated a synthetic data set to work with. And this is basically just how I did it. I said I want 1000 per group. I know the mean I know the standard deviation. And then I can generate numbers that that give the exact same normal distribution as is described in the paper. And then we can fully look at those because most people don't give you all that data, you have to kind of find ways to extract it. So as my first example, this is FTO or fat mass and obesity associated protein. It is considered to be the common snip that is most associated with obesity and being overweight. And you'll see that per a copy. So this is a switch from a thymine to an adenine per a copy. Your BMI increases by 0.3. So I'm at I'm here in the middle. And what I've done is using the per allele effect, I've plotted some normal distributions, assuming that TT here, like this is the there's no effect. And then, you know, how much of an effect how much of a change in BMI do we get as we go up? And you'll see the average, this middle line does increase. But the likelihood of me, somebody who's AT, just having no effect or being or having a BMI below that of some in the TT group is 46%. It's basically a coin toss. And the same thing if I have two of those alleles is 42%. So even though it says on average, yes, BMI might increase as you know, as you increase the number of FTO snips that you have. In reality, the data is so variable, that doesn't really tell you anything about your own obesity risk. To kind of just show you how I did that. Again, in a very similar process, there are free tools online where if you can put in the mean and the standard deviation of the data. So here, this is the the mean effect per allele. Here's the standard deviation. And then I say, what's the likelihood that I'm going to be below zero? So no effect of these alleles. And it's 46%. So again, this is something that anybody can do with data if they're really trying to understand it. So if we then take that data and try and put it into action, this is data from a cohort in Finland. The reason why I've picked that is because it's the largest data set on adults that's who are close to me, they are called mainly Caucasian, they're 31 years old. So I know that some of the confounders have been removed. And what I've done is I've again plotted those bell curves. So you see TTAT, here's me in the middle, AA. And then I've got the likelihood of a BMI below 25. So just being a normal weight in each group. And you see that the increased risk of being overweight is only 3% in people with my genotype. And even people who have the best genotype over here, 41% of those are still overweight. So the effect of the genetics is far smaller than just the baseline effect that you're seeing in the population. So if we then try and figure out the effect that the gene is having, what we can do is plot those data that I had against the number of A allele copies that people have. And you see that at least in this data set, here the R squared, which basically tells you how much each step changes the slope of this line. Your FTO genotype explains about 0.2% of your BMI. That's like whether you woke up and had a glass of water in the morning or not. Okay, so that's only one gene. So we've now, or people have looked at all the different SNPs that most affect obesity risk. So this is eight total SNPs. They created a total genetic score. They applied it to the Epic Norfolk. There's a cohort of people in the UK. And just to orient you on this graph, here on the left y-axis, you have the number of people in each group. You have the the the obesity risk total risk score based on genetics, sort of increase along the x-axis here. And then this is the average BMI per risk per risk group. And you see, okay, there's you know, there's kind of a nice linear effect there. It seems to BMI seems to go up as genetic risk increases. However, probably the most important thing to note here is that most people have quite a high risk genetically. It's actually quite rare to have a very low risk. And it's also rare to have a very, very high risk. It's just like most people have some risk of being obese, which I guess you're not really surprised by. So if we plot things in a similar way, so for each genetic score, I've recreated the bell curve, I've added lines here to show who's obese, who's overweight. And you see, yes, if you go from the lowest risk to the highest risk, BMI increases on average by about 1.4. And it increases the risk of obesity from 6.9% to 20%. That sounds like quite a lot, but you'll still see this huge variability, like regardless of your genetics, you could be anywhere on this scale. And the population, regardless of genetics, more than 50% of them are overweight. And so this starts to tell you about the data sets that people are using to analyze these genetics, and maybe not the data sets that are that relevant to us. So we have done the same thing here. I've plotted the risks all here, assuming a linear effect, which is what the people in the paper did. And your obesity genetic score, based on all of the genes that most affect the risk of obesity, explains like 2% of your BMI. That's whether you went to the bathroom before you weighed yourself in the morning or not. So if you're all worried about your genetic obesity risk, you've put your 23andMe data into some kind of calculator. It says you have an X% increased risk of being obese. The most important thing is just don't live a post-war Western lifestyle. This is a really nice paper which looked at those those three different FTO genotypes. And it looked at people born before or after, essentially the middle of the Second World War. So these are people with a low risk obesity genotype. They're pretty much the same. This is my genotype in the middle here. And this is actually where we see the biggest increase in risk between these two supposedly. But almost all of that happened in people born after 1942. So before 1942, that's the blue line down here. FTO wasn't associated with obesity or overweight at all. So just think about the environment that you can create for yourself. Don't eat like the average American 12 year old and you might be okay. So again, if you're worried about your obesity risk, just don't live entirely like a post-war American. And just moving frequently. So almost every study that's looked at activity on genetic risk of obesity shows that an hour a day of activity. And that's just like if you stand for more than like up to an hour a day. If you have a standing job, this counts. So it's not even like you have to go to the gym or run 10 miles every day. Just a small amount makes these effects almost completely negligible. And this is the same advice regardless of genetics. So people have done a similar thing for diabetes. Obviously blood glucose regulation is super important for long-term health. This is kind of a jump to a busy graph. So I'll kind of walk you through it. Here is a glucose genetic score. It's constructed exactly the same way as the obesity genetic score. The two populations on the right here are two hunter-gatherer populations, the Tuca Sentence and Catarvons, where they published the mean and standard deviation of the blood glucose levels that they have. And then I've added a disease risk line. This is at 90 milligrams per deciliter, fasting blood glucose, above which is when we start to see increased risk of cardiovascular disease, obviously type 2 diabetes and decreased mortality. And you'll see that, oh and I've also added a couple of lines here. So there are, I could find four hunter-gatherer populations where they gave just the average blood glucose and they seem to hang out somewhere between that, you know, the high 60s and the mid 80s. They also have just a, you know, maybe a couple of percent chance of being over that 90 milligrams per deciliter line. So you'll see this is data from the framing in population here in the US that the average genetics, which again is right in the middle here, you have more than 70 percent increased risk of having elevated fasting blood glucose that's entirely driven by the environment. You know, we know that hunter-gatherer populations aren't just superhuman blood glucose regulators, right? You put them in the western environment, they'll get just as diabetic as the rest of us. Importantly, compared to that massive increase, going from the best genetic score to the worst genetic score only increases your risk by, I put 13 percent, but it's actually 23 percent. But it's still, you know, it's a minor component. These, these are also very rare genotypes, most people are hanging out in the middle here. So if I plot that same, same graph that I showed you earlier, your glucose genetic score against 16 different SNPs that are all significantly associated with elevated fasting blood glucose, that determines five percent of your blood glucose level. Five percent is more, sorry, is less than the error of most repeated measurements on standard home glucose tests. So just repeating your blood glucose test again and again, that's going to be more variable than the effects of your genetics. And this is where I think it's really important to remember who the comparison group is. So when we're taking data like NHANES or Framium, there was a recent paper that just came out that said that, or showed that 82.4 percent of American adults have suboptimal metabolic health. It's actually probably much more than that because, for instance, their blood glucose cut off for suboptimal metabolic health was 100 milligrams a deciliter. We already know that's pretty high. In the U.S., 11 percent of adults 45 to 64 have type 2 diabetes, 20 percent over 65. And if you then compare that to say another hunter-gatherer population, this is the Bolivian semenae that this paper just came out last year, they have a 0 percent risk of diabetes. So if you live over there and your genetics increase your risk by 67 percent, 67 percent above 0 percent is still 0 percent. So these problems are entirely driven by the environment. So that kind of briefly brings me back to that question about the fat versus carb genotype. And it's really interesting because if you look, this could be a whole talk it all in of itself, but basically the studies that show those kinds of things, they measure, they look at your SNPs, they ask you what you eat and then they try and infer something out of that. And we know that nutritional epidemiology is so flawed that it's essentially useless. And now we've just added a layer of complexity which is genotyping and it doesn't really show anything. And there is no single study that has randomized people to any kind of diet based on the genetics and seen any difference in health outcomes. Probably the best we have so far is the diet fits trial run by Chris Gardner. And you see that he sort of they randomized people to either a low carbohydrate or high carbohydrate diet. Improving diet quality was probably the most important thing and then they sort of iterated over time. But then they went back and looked at people's genetics and how much weight they lost. And so you have low fat genotype where they got low fat or low carb diet, same for low carb genotype and then people who were sort of mixed. And you just see that like weight loss is essentially the same in every group regardless of diet, regardless of genotype. So moving on from the body we'll move up to the head. You may have heard of the warrior gene. This is actually the SNP causes a change in amino acid in the COMT gene from methionine to a valine. If you have two methionines you are classified as a warrior. If you have two valines you're classified as a warrior which is me. That means that you have high COMT activity. COMT breaks down dopamine. So that means I have lower dopamine and probably look like this in my mind. That's what I'm imagining. But if I read up people will tell me that because because of that I have a lower IQ. I have lower executive function but maybe I'm good in the fight. So I'm probably useful for something. So where does that information come from? And this is super interesting. I went and this is the paper that most people cite. And they took deceased people's brains, 105 of them, and they extracted their COMT enzymes and looked at their genotype. And then they looked at how much activity those enzymes had in a test tube. So you'll see here, this is me, this is the warrior, this is the warrior and you see that supposedly we have a 38% increase in just baseline activity. What's really interesting is that nowhere in the paper does it tell you what the error bars are in the graph. So I have no idea how variable this data is. If I'm charitable, I'll say okay, it's a standard deviation and then this is what the bell curves look like. And so I'm here, I'm the warrior, I actually have a more than 99% chance of my COMT being more active than the average warrior. And I think, okay, that makes sense. However, I'm pretty sure that these bars are standard error of the mean just because they're smaller in groups with higher numbers and just knowing a little bit about trying to do these kind of experiments in the lab, these are deceased samples, they're really hard to get, really hard to process, you then have to do the sample, you do the experiments in a test tube. The data is probably a lot more variable than it is here. So it probably looks like this. Now, if you're paying attention, you'll see that when I'm trying to recreate these bell curves from the data in the paper, my random number generator wants there to be negative values, which is like impossible, you can't have negative activity of an enzyme. So that means that this data is not normally distributed. And it was hugely variable, which is very likely based on the kind of samples that there are. We just see how much overlap now there is between those. And so like when anybody's telling me you have 38% more CMT activity in your dorsal lateral prefrontal cortex, that just doesn't mean anything. These people don't know anything about the data that they're talking about, unfortunately. All right, so this is just like some in vitro scientific like bugbear of mine. So like what about something that you might care about, like cognitive processing? So I'm told lower executive function. And people have done a whole load of battery of tests in people with these CMT snips. And the performance is usually very variable, as you might expect. But the data I have here is from one paper where they had a large number of people, I think it was about 500, they put them through a big battery of tests, and then they looked at their performance. And most tests didn't really make any difference. But in the letter number test, there was a significant effect of genotype. So I've plotted that data. The letter number test is basically you give people a random list of letters and numbers, and then they have to recite them back in alphabetical and numerical order. So if I just plot those, plot those data, you'll see that, you know, even though there is a difference, like if I'm a warrior, I do have a lower score on average. About a third of people who are in these two middle groups would perform better than the average, the average warrior, and about a quarter of the people who are who are warriors who should perform better will perform worse than the average people with the fastest genotype. So you could be anywhere on this, you could be anywhere on this scale and you're never going to be able to predict that. So again, I plotted a similar graph and I showed that your CMT genotype explains about 4% of your performance in these tests. And so I think that gets really, you know, a good place to talk about what a gene does, right? And so most people think, you know, so you have a CMT gene that makes your CMT enzyme, and that turns your dopamine into HVA. There's a couple of steps in there, but that sort of breaks down dopamine. But, you know, there are transcription factors which turn on that gene. And those transcription factors are essentially controlled by the environment, the internal environment in the cell, outside the cell, outside of the body. And then you've got to make dopamine in the first place, right? So from dopamine, you need tyrosine, you need some cofactors, you know, some vitamins, and they also come from the environment. And then, you know, if you want your CMT enzyme to work, you need esudenosyl methionine, SAME, and that means your methylation cycle needs to work, and that's, you know, largely determined by the environment. And then, you know, your dopamine acts as a receptor, and then it creates a dopamine signal, and then that tells your body, well, hang on a second, maybe I have too much, too little dopamine, and then that alters the transcription factors. So what determines the amount of dopamine in my brain is the environment, all these feedback loops, and it's not just an isolated enzyme that somebody purified and looked at the activity in the test tube. So if you're worried about your cognitive function, worried about your CMT, all the usual stuff, maintain robust circadian rhythms, move frequently, eat real food, reduce chronic stress. And for me, that's obviously, I should stop worrying about being a warrior. And so then this quote really sat with me, this is from Robert Sapolsky in his excellent book, Behave, which is that having the warrior gene probably has less of an effect on your behavior than does believing that you have it. And this is born out in multiple different studies. So here's a really interesting study where they took people, they made them run on a treadmill, do an endurance test, and then they, they did a genetic test and they told them, you either do or don't have the genotype for good endurance, right? But they did that regardless of their actual genotype. The people who were told they had the bad genotype got worse in the next test. The people who were told they had the good genotype stayed exactly the same. So, and that's regardless of what the genotype actually was. So finally, I know you're all waiting for this one, MTHFR, this is probably the most talked about SNP or one that we hear about most frequently. There are two that are normally talked about in combination and they have, they have almost additive effects on the function of your MTHFR gene. And luckily for me, I'm homozygous for both. And so I went into an online calculator which said that if you have reduced MTHFR function, you need more choline. I thought that's super, super interesting. I'd like to know more about that. So it's, you know, choline is an important source of methyl groups. I have a 53% reduction in my MTHFR activity. And also because of some other SNPs, actually, I've reduced my ability to produce methyl folate by about 65%. And therefore I need 1,088 milligrams of choline per day. That's the equivalent of eight egg yolks, a pound of liver or two pounds of salmon. And like I'm a pretty big guy and I eat quite a lot, but like I just, I just can't get on board with that. Like is, do my genes really tell me, with no other information, do they tell me that I need to eat more choline that I probably realistically would have gotten, you know, like out in the wild? And that sort of really makes me ask some questions. So at the top we have here, this is the data on MTHFR activity like we had for the COMT with the different genotypes. So, so here's the six, 7,7 CT. And this is sort of like decreasing function. And here's the 1,2988 with decreasing function. And they can kind of overlap. These ones here are super, super rare. But here's me in the middle of 53.2 52.3% loss of MTHFR activity. And what's really nice about all of this is that from across all of these combinations of genotypes, we have like from a 0% loss of activity down to a 72.5% loss of activity. So we can look across that whole range and how that affects our physiology. So I did a similar thing. I plotted the bell curves based on MTHFR activity. You know, here's me down here. I have a 100% chance of having my MTHFR be less active than the wild type these guys over here. And so like, I was actually quite shocked by that. I thought it wouldn't be that much. But you'll see that these guys down here, the 6,7,7 TT, they're even worse off and we'll have to look at them separately because they're a bit I'm being told to stop. But you guys don't want me to stop, right? No, okay. So you might have seen a picture like this talking to you about MTHFR. So that it's here and one way to kind of check the way, you know, how the whole system is looked the working is to look at home assisting. So you need MTHFR to make methylfolate, to recycle home assisting. And if you don't have enough methylfolate, you can use choline to kind of jump in and do that. That's why choline requirements increase if your MTHFR gene doesn't work. So let's look at home assisting levels across these different genotypes. So here I've added a line, a home assisting of 10. That's where, again, disease risk starts to increase probably significantly. So here's me over here. In the people with the best. So this is the same paper that looked at the activity of the enzymes and people with the best activity, they still have an 83% chance of having a home assisting over 10. And that's just increased by 8% in me. And so if we're trying to figure out how all of this works, this here might be the effect of your genes, but this is the effect of the environment. And again, remember the comparison group. So I showed you that picture. I had all those different B6 coding, all those different things. If you're comparing to the average American who doesn't eat enough folate B12 from bioavailable sources, then you might think that these are really important, but in reality, particularly for the people in this room, that's not going to be the case. So in a very similar way, you know, if I look across the most common genotypes, more than 85% of people, your MTHFR activity down to 48% or 52% loss, this is me here, it explains about 1% of your home assisting. That's again, like just lab test error. But then let's add in these guys here. This is the 677TT, probably one of the SNPs that people talk about, the most worried about it. And you'll see the bell curve isn't a normal shape. Again, my random number generator tried to make some negative values here, which I cut off. So that tells me this data is not normally distributed. These are not a normal population. They're going to be hugely variable depending on a lot of normal things. And these guys are clearly different. We can't just lump them in with everybody else, which is what tends to be done. So this is what people do. Here I've added them to this graph. All of a sudden, your MTHFR activity explains 13% of your home assisting. But that's just because these guys have completely skewed that data. So if you're existing here, in reality, your MTHFR doesn't really make any difference when it comes to home assisting. But so what happened in that coding calculator was, here, these are people with the best MTHFR. Here are people with the worst, this special case. They need twice as much. And then I was given a score of like 65% last. So you just kind of like read off the graph and I need 1,000 milligrams of coding. So that assumes a linear effect between the two. But I've already shown you that the effect isn't linear. Those C677TT guys skew everything. And then when I looked to the references for the coding requirements in that C677TT group, it was in 13 folate compromised Mexican American men. And just starting with ethnicity, we know ethnicity is so important when it comes to how these genes affect us. So like FTO, associated with obesity and Caucasians, but it isn't in African Americans. So knowing nothing else about me that you can actually then try and interpret something, tell me how much coding I need. Like, I'm out. That's enough. I'm sorry. But I did think, you know, is others C677TT guys like a special case, should they be measuring their MTH afar and worrying about it? And it's really interesting because when you look at home assisting that's elevated in those people, it's almost always associated with nutrient deficiency, B12 folate and particularly riboflavin. So this is a study that Alex Leif has published or publicized a lot recently and here you see these are the C677TT guys. You give them 1.6 milligrams of riboflavin which isn't very much a day and they decrease their home assisting by 20%. That's because this mutation affects the ability of the enzyme to bind riboflavin. So just give a bit more riboflavin and everything goes away. So are you worried about your MTH afar? Measure your phenotype, not your genotype, home assisting folate, B12 status, eat foods that are rich and all those things that support those cycles and all of those are necessary regardless of your genetics if you want to optimize your health. If you're still worried, take a few grams of glycine and creatine, they support everything, take a few milligrams of riboflavin and just stop worrying about it. All of these things are super low risk and high potential benefit. You can't, it's very difficult if you just follow that to take too much. All right, so just summarize, I know that Stephanie is probably having palpitations right now. So a few other reasons why I think we should reconsider the amount of importance we put on genetics. The first thing is that we don't understand how proteins interact. There's a paper that just came out that they analyzed across the genome, proteins and different combinations and how they might interact. They discovered several hundred protein combinations that would interact in a way that we've never seen before. We also are starting to see that all genes affect all other genes. So in a specific disease associated cell, every gene that's being transcribed in there is affecting every other gene and we just do not understand those interactions well enough yet. We also don't even know the scope of what we're looking at. So there was a paper that just came out that looked at all the different transcriptionally distinct cells in a mouse brain. So that's cells who were trans who grouped together transcribing different types of genes or in different amounts. There are five hundred and sixty five different types of distinct cell in the mouse brain. We don't even know what that is for the human brain. So if you're telling me that my COMT is determining how much dopamine is my prefrontal cortex, but I don't even know how many different cell types that are in the brain. And I'm a neuroscientist. Like it doesn't make any sense. Also, if you're a practitioner, what's really important is there have been multiple studies that have showed that just telling your clients or patients about their genetics does not change their behavior at all. So we really, it's not really something we can hang our hats on. When we think about how we think about genetics, it's really interesting that telling people about their genes if they have bad snips causes a negative effect on health and performance, but telling them that they have the good version doesn't have a beneficial effect. So that we're automatically primed for it to cause a negative effect on our health. And I think that's partly because of the way we talk about it. So genes are always normal. So MTHFR, with 100% function, that's the normal gene. Everybody else is bad, right? And so that automatically you're going to get this noceba effect. Oh, I'm a poor methylator. Oh, my MTHFR doesn't work, even though that's normal. And then we're also going to start to create this worry of or what if I'm not doing something that's in line with my genetics when we have no real evidence of what that might be. And I think this is kind of this is coming to light becomes a bigger part of us because we've moved away from these collectivist hunter-gatherer cultures into this individualist American dream, every man from self-type culture. And that's automatically going to increase the amount of importance that we put on ourselves, the things within us. I'm special. When in reality, almost everything to do with your health, chronic disease is determined by environment. So my question is whether actually just doing your 23 and me analysis in the first place is an evolutionary mismatch. So just very briefly, I think these reductionist S&P or SNP approaches are problematic because if you don't know your phenotype, you can't act on them in isolation. And if you do know the phenotype, then you'd make most interventions based on that rather than on the genetics. If you have the phenotype and the SNP, you don't know if the SNP is causative. And for most SNPs, we don't have any specific interventions to affect those. So finally, things to take away. As a user, really important to ask for evidence of the recommendations you're getting based on your genetics. Like, look at those studies. How much for those participants like me? Ethnicity, super important. Socioeconomic status, those of you in this room are really lucky to be here and you live in a country that has some of the worst health disparities. And if you're looking at the average American population, the people with low socioeconomic status are much greater risk of all the things that you might worry about. Childhood environment, if you think about the neurotransmitter genes, they're only associated with worse mental health outcomes in people who experience childhood trauma. So all of this is super important. Then just the nutrient and physiological status of these people, is that similar to you? How variable is this data? And most importantly, what's the likelihood that my SNP will have no overall effect? Because there's actually a number where the protein is no overall effect. As a practitioner, do some of these calculations before you start recommending things to people. I'll put a step-by-step guide on my website. There's going to be a lot of people offering tests now with thousands and thousands of SNPs. Our test is better because it has this X-Many thousand SNPs more. I've gone over the SNPs that we probably understand the best. The vast majority of them we have no data on to even have this discussion. And then just finally, as a human, just know that genetic-based interventions are far from understood. Disease risk and gene function are clearly dominated by the environment. And if you're watching this talk, you're already so far ahead of most people that there's studies that are in the studies that you're trying to compare yourself to. So, you know, that really needs to be born in mind. Finally, just like to thank everybody that I work with and thank you for listening and there probably isn't any time for questions, but I'll take some if I can.