 recording now. So welcome back everyone. So if you want to design a primer for the human myoglobin gene and you for example want to amplify the coding sequence between 153 and 464 base pairs then of course the first thing that you need to do is get the sequence. So what you do then is you copy paste the sequence into primer 3 and then you say I want to pick a left primer I want to pick a right primer and then you want to specify which you want to target and so you want to target a certain region of the gene and then after you specified which target you wanted to have you say pick primers. What it will show you is the output so the output that it will show you it will show you the region where it picked the primer so this is the first primer the forward primer it will show you the target sequence highlighted with stars and it will show you the reverse primer and that will be the other way. Alright summary far better than my machines many programs are available and the example that I showed you of primer 3 I will actually do a live primer 3. So here let's go back to the cow sequence that we had so this is the cow sequence that we had the snip here is in the middle right so it's at 500 base pairs after and 500 base pairs before so we just copy paste the sequence so let's copy the sequence we go to primer 3 and we just paste the sequence in so there's two ways of specifying where you want to have your target you can use the target field however you can also just highlight the sequence using the square brackets so because can I make the box a little bit bigger yeah I can right so here we see our sequence so in this case we know where the target is exactly so it's better to use the target here so the target is specified as 50 comma 2 which means that you require the primer to surround the two bases as position 50 and 51 so in this case our target is at 500 and the target is no I will say the target is at 499 and we want to span three base pairs so we want to have the primer located at or we want to have the primer spanning 499 500 and 501 so just that the primer is not like stuck directly to the to the snip I want to have a left primer and a right primer and I'm just going to say pick my primers and in this case here we see the target sequence so the snip is in the middle so that's the G here and we have the forward primer being chosen and the reverse primer it will tell us then that the left primer starts here the length you see that the TM is relatively close to each other so it's it's slightly less than 2 degrees Celsius which is perfectly fine there's no chance of a hairpin there's no chance of dimerization with itself and the three prime is also perfectly fine if these numbers are not zero these these three numbers here then there is a chance that there will be a primer that there will be a dimer or that it will fall back on itself but in this case these primers look perfectly fine so this is the sequence that we need to order for the forward primer and this is the sequence that we need to order for the reverse primer and then here we would amplify this piece and it will tell us that we would get a product of 231 base pairs so we when we when we would do the reaction we would put it on a gel and then it should be at around 231 we should see a little band and this band is the band that we want all right so that that's it for primary design like it's not it's not magic it's just that you have to be very careful and of course make sure that you always mask your sequence using either using ensemble or using the repeat mask or tool to just get rid of the repeats because there if you would not have repeat mask it then it might have chosen a primer in the wrong region all right so that's it for the primer designs are there any questions because if there are no questions then I would like to spend the remaining 45 minutes talking about my PhD thesis about the CTL mapping in relationship to the QTL mapping part I'm just gonna wait for a little bit to see if anyone has a question on how to design primers if you want to learn more about guest mirrors or if you want to learn more about semi-universal primers then just just ask if not then let's go to the other PowerPoint then good so this is more or less where we stopped last week so last week we I talked a lot about quantitative trait locus mapping right so about the standard QTL mapping so a QTL a quantitative trait locus is a section of the DNA which is called the locus which is associated with a variation in a phenotype so phenotype here is always a quantitative trait not a qualitative trait because otherwise we would be using another quantitative trait mapping so I showed you this one right so this this slide before and so to detect the QTL in a population we have to measure the genotypes so in this case we have AB genotypes and for example we can use SNPs or AFLPs or ROFLPs and we have to have our phenotype of interest measured for example the tail length of an animal or the yield or the amount of protein in the milk so here for example we have a whole bunch of mice so 1 to 9 and then this will continue up until like 3 400 if you have 3 400 mice measured all of these have a have a phenotype and all of these mice come with their genome being part or either coming either from the mother or coming from the father and so at a certain genetic marker have we compute the difference and we want to see if the mean of the a group is different from the mean of the B group and then after we see that there is a difference we also want to know how likely that difference is and so again the nice little animation that just goes through all of the different markers on the genome and for every marker you will get an effect size and it will tell you if the mean of the a group is smaller than the B group or if it's larger than the B group so just basic QTL mapping just go through the genome and marker by marker scan it and write down the difference we use statistics to assess the association of the markers and we can use things like t-tests or an OVA or regression and this really depends on how your genotypes are structured and if your phenotype is a normally distributed phenotype or if it's different so and there's had the model that you have to use is dependent on the phenotype distribution and the amount of groups that you have in your genotypes and we always express it as a minus log 10 of the p-value right so you plot that across the genome we already saw this one before and here in R we can for example use our QTL one of these packages which is made for QTL mapping in R to do the QTL mapping and so if you want to produce this picture and we can go to we can load the library so we say library QTL we load this data set which is the hypertension data set in my so we just say data hyper and then we just plot and the function that scans for QTL is called scan one so just scan one the hyper data set and then we just plot that and then you get this picture so QTL mapping comes with some severe limitations one of the main things is that it only considers a single phenotype at a time right because we're just looking at a single phenotype and at each marker we split into two groups or into three groups and then we look to see if there's a difference in the mean of this phenotype and so we only consider a single phenotype one of the other limitations is that the phenotype needs to show significant differences because if a phenotype is only varying by a little bit right if you look at for example the amount of the amount of arms of a certain of of a human right so almost every human has like two arms and there's no big variation there at least not genetically and well it sometimes happened that people get born with like three arms or no arms but it like it is very hard to find a genetic location which is involved in things which do not vary right that's one of the limitations of QTL mapping so hey it's it's if there's no variation you cannot associate it with a location in the genome and one of the other problems is that if two phenotypes are like very correlated to each other then their QTL profiles look exactly the same because mapping the one phenotype is similar to mapping the other one because they are 99% more or less correlated to each other and so this is one of these reasons why QTL mapping is sometimes not the best method to use so when I was doing my PhD I thought about how can you kind of circumvent this these limitations right these limitations are very detrimental to the method in a way because hey you you want to not just consider a single phenotype you want to look at multiple phenotypes or the relationship between them hey if a phenotype doesn't really show a lot of significant differences it might still be that there is enough difference that combined with another phenotype you could find a regulator or a locus which is controlling your phenotype so imagine two length phenotypes in a plant population and in plants it's very common that you have for example that your susceptibility to infection and the amount of produce that you make so your yield and your susceptibility to infection is generally highly correlated with each other right because if I improve the yield of the plant automatically this plant starts becoming more susceptible to infection has so bigger yield is higher susceptibility and in plants we call this the yield plateau so if you very but if you look at for example the selection in wheat fields from like 1892 to 2000 and then you see had that the first part there was no real improvement and had then we had pedigree selection so hey you see that the average goes up a little bit and then at a certain point like in 1950 we started doing scientific breeding and we started to use like we started to use things like more or less association and cutiel mapping right and then you see that the yield of of of wheat actually goes up quite tremendously but you see that around since the 1990s we don't see a massive increase anymore because phenotypes are linked to the yield phenotype when we are selecting for higher yield we are making plants which are more susceptible to infection or which are not as which cannot protect themselves against like wind and these kinds of things anymore yeah because the weed becomes so big that when there's a little gust of wind the thing falls over and the plant dies and so when we started doing like the scientific breeding approach using either pedigrees or using QTL mapping information we see a massive increase but this increase is leveling off especially in the last like 10 years we haven't been able to improve plants as much as we could in like the 20 years or 30 years before that so this is called the yield plateau so the method that I developed is called correlated trade locus mapping based on on QTL so I define a correlated trade locus or in my thesis I define the correlated trade locus as a locus is a section of DNA which is associated not with a difference within a single phenotype with between the means but which is associated with the difference in correlation between two phenotypes so how do we then do that right because normally we would just go at a certain marker we look at our phenotype and we split it into two groups and then we look to mean in the one group and the mean in the other group so CTL mapping is very very similar to QTL mapping but it is multi-phenotype and because we're now not using a single phenotype we're using a pair of phenotypes so what do we do we identify genetic regions where there is a difference in the phenotype phenotype correlation structure and that is conditional on the genotype of the marker in one of the advantages of CTL mapping similar to QTL mapping is that it's unbiased and it's data driven so you you you don't have to have any prior information of course you have to have measured two or more phenotypes but normally when you do an experiment you're not measuring a single phenotype you're already measuring multiple phenotypes right if I'm if I'm doing if I'm a plant researcher I'm measuring like the yield of the plant the number of leaves the height and all of these things the susceptibility if I'm a fish researcher I'm also measuring more things than just the single phenotype that I'm interested in so applications of CTL mapping are that you can have classical selection in breeding to improve economically interested phenotype because we can get rid of this yield plateau if we could break the correlation between the susceptibility and the improvement in yield right so we could select for beneficial CTL locus similar to how we could select for similar QTL load size right so for plants and yields have we would want to select for markers where the correlation between yield and susceptibility is low and this will allow us to break this linkage between the different phenotypes and of course we need to combine this with QTL data and if we combine it with QTL data CTL mapping also allows us to build phenotype phenotype connection network so have which phenotype is is connected to other phenotypes so hey in this yield plateau if we have both phenotypes yield and susceptibility will show very similar QTL profiles because of their overall correlation and because yield is is highly correlated with susceptibility when we scan for yield we get a QTL profile when we scan for susceptibility we get a QTL profile which looks very similar but CTL mapping tries to find load size in the genome where this correlation between yield and susceptibility is lost and so we can use this CTL information to break the link or the correlation between yield and susceptibility allowing us to have like one generation of selecting for individuals which do not show correlation between yield and susceptibility if we do that for one generation then in the next generation we can select for high yields again but now increasing the yield will not lead to an increase in susceptibility because in the previous generation we already broken that link we already broken that that combination so a question to you guys imagine that for example phenotype a here is the yield of the plant phenotype b is a susceptibility and now we're looking at a single marker in the genome and we see this picture so we see that in the AA genotype they are strongly correlated and in the BB genotype and there's really no correlation in high yield individuals have so individuals which have a high yield have a susceptibility which is more or less medium or they have a susceptibility which is high so my question to you if if I am a plant breeder and I see this picture of a certain marker right which which genotype should I put into the ground for the next generation should I put individuals in the ground who have a strong correlation which have the AA genotype or should I put individuals in there which have the BB genotype which have low correlation and I'm just going to wait for an answer we can actually do I figured out that we can actually do a how does twitch call that twitch has a prediction function so we can actually start a prediction and then you guys can kind of bet on it using your channel points and then the winner gets the channel points from the people that got it wrong we should do that more often to kind of but my question to you guys is if I'm a plant breeder and I see this picture which plant should I breed should I breed the AA genotype plants or should I be breed the BB genotype plants no answer in chat so I'm just going to assume that everyone's sleeping so of course in this case you want to breed the BB genotype individuals because the BB genotype individuals do not show the relationship had they do not have the correlation had because if we would breed the AA genotype we would not lose the correlation between yield and susceptibility all right so you can use I thought we were betting on it ah damn it then say that like come on people you were waiting for me to start the prediction we will do that next time next time that we have a question I will I will make sure that we start a prediction on it but but you can use the correlated trade locus mapping technique to select genotypes which unlink the two phenotypes that you don't want to have linked to each other and so the next generation now shows a decreased correlation between the phenotypes which allows you to select for high yield again without increasing susceptibility so it's the same as so and afterwards so after one generation of breaking you can then select individuals again based on standard QTL information because now all of a sudden the QTL profile for yield will look different for susceptibility all right so the methodology I think that it's it's interesting to just see how we do that so to explain the method I using recombinant in red lines right because then we only have two genotype groups and we don't have to deal with like four or five or heterozygous and these kinds of things but the the CTL package that I wrote which is available in our can handle any type of cross so it doesn't matter if your species has two chromosomes or if it has four chromosomes it doesn't matter how you cross them if you have like homozygous AA heterozygous homozygous BB or if you have four load site and you can have everything like four A's three A's and a B two A's two B's so the package deals with that but recombinant in red line so in this example I will assume that we have four different phenotypes that we measured and we will be just looking at six genetic markers and at these markers there's only two possible alleles the AA allele or the BB allele so how does it work we select a phenotype so P1 for example and then we select a genetic marker in the genome so here in this picture we're only looking at a single phenotype and at a single marker so what do we do we split the individuals in two groups by genotype so we have AA individuals and we have BB individuals so and what do we now do is we calculate the correlation of both the AA and the BB genotypes of P1 times all the other phenotypes and then we get a correlation vector right so how we do the correlation of phenotype one versus the other one so Px in AA so we just take only the individuals which are AA at marker one and then we calculate the correlation of course the correlation of P1 to itself is always one that's just the way that correlation works yeah because that that but for example the correlation of of P1 to P2 in the AA group is 0.1 the correlation of P1 versus P2 in the BB group is 0.2 right and we can we just do that for all the phenotypes that we have so how we get two vectors one vector is the correlation of P1 versus the other phenotypes in the AA group and we have P1 versus all the other phenotypes in the BB group so what we observe now here is that when we calculate QTL effect size we take the difference in mean between A and B if we talk about CTL effect size we take the difference in correlation between the AA group and the BB group and so we make a difference vector which we just say well we just subtract this one from this one so have 1 minus 1 is 0 0.1 minus 0.2 is 0.1 so it's the difference it's the absolute difference between the correlation in the groups and what we see now is that there's something interesting because we see that at this marker there is a big difference between the correlation of P1 versus P4 because in the AA group it's 0.8 in the BB group it's only 0.1 so we see that there's a 0.7 difference in correlation so now what we do and this is the big trick we now do this not for one marker but we do this for all markers in the genome right so here we see the result from the last slide right so and now we just we just took we took this vector that we calculated and we just flop it's on its side so we just make this the first we make this the first marker right and then we do it for marker two marker three marker four marker five and marker six right so we have six markers four phenotypes and then what we see happening is when we repeat this calculation for our for every genetic marker we get multiple vectors and every vector if you have multiple vectors we can combine them in a single matrix and then what you will see is that there's also linkage so hem marker one and marker two are very close to each other and so the correlation difference for P1 versus P4 at this marker is 0.7 at marker two it's 0.6 and then you see it go down quite slowly and for P1 with P3 we see this as well the thing is is that you don't get just one matrix right we're doing this for P1 versus all the other ones but we're also doing it for P2 versus the other ones and for P3 versus the other ones so we get not one matrix but we get four of these matrices and of course if you map P1 against P1 the difference is always zero right because the correlation is one in the AA group one in the BB group so the difference between one and one will always be zero so based on the matrix right you can see which phenotype you mapped if you take P2 and map against the other phenotypes then this one would be zero if you take P3 map it against the other then these would be zero then these would be zero all right so now we have to do permutation right because we if you have a difference in mean if you have two groups right and you have a difference between the mean then you can just do a basic t-test but we cannot do a t-test here we we we cannot t-test based on correlation values right because we don't have a group of individuals anymore so at this point we now have to start repeating our analysis so we have to use permutation to find our significance right because here we see our effect size matrix this is the effects the CTL effect size of P1 versus the other phenotypes at each of the marker but we don't know if this 0.7 is really a significant difference right so we need to just do permutation so we break the link between the genotypes and the phenotypes right so we sign genotypes to individuals at random and we redo the whole analysis and we remember the maximum observed score so the maximum observed difference in the correlation and then we make a distribution out of these 10,000 scores and we find the 5% and 1% threshold values for significant just like in QTL mapping but now of course the method is much more involved because we have to do correlation calculations a lot which are not they are more expensive than calculating the mean and so we use 10,000 permutations to assign the significance you could also use a direct calculation of P values using mathematics so if we if we if you use the CTL package that I published for R then it has a direct calculation of P values as well as a permutation approach and we do the same thing again so we convert the differences to P values and then we convert these P values to lot scores by using the minus log 10 P value just like in QTL mapping so what do we do we convert these to to minus log 10 P values so when we perform a QTL mapping for P1 and we get I think this should say when performing a CTL mapping for P1 we get four vectors of lot scores from CTL mapping and we get one vector of lot scores from QTL mapping right so here we see the different vectors so for for P1 against P1 the lot score is always zero right because the difference is zero the lot score is also zero but we now see for example that we get scores of 7.1, 6.2, 6.2 and 5.1 in the likelihood that there is a significance difference in correlation and compared to the information that you get from QTL mapping right because we also QTL mapped P1 across the genome hey you get like a QTL profile so we have a QTL profile where for each marker we have a single score but for CTL mapping for each marker you get an n amount of scores where n is the number of phenotypes that you scan so how do you then visualize it so this picture is a little bit annoying because it doesn't really show below but you get a QTL score of P1 so I always plot those at the top so this is the the the same phenotype as that we did during the assignments last time so you see here that there are two peaks one on chromosome 4 one on chromosome 5 but what you see in here in the bottom is that you see three different lines or four different lines like one of the lines is just zero zero zero right so in color you see the four different lines and so you see that there's a little bit of correlation lost with the red phenotype which is the third one and you see here in blue you see that there's a big correlation loss with the one on on chromosome 5 and the scale here is a little bit off but they are significant peaks if you want to call it like that all right so that was the thing that I wanted to tell you about CTL mapping so CTL mapping is very similar to QTL mapping but instead of taking taking one phenotype and doing the mapping and what we do is we take two phenotypes or a pair of phenotypes or multiple phenotypes and we just say well we want to know one phenotype versus the other one and so in CTL mapping your output is not just a single QTL profile it is a single QTL profile plus a correlation profile for your phenotype that you selected versus all the other phenotypes in your data set and then you can do nifty stuff with that because the information here allows you because in some cases if your phenotype of interest does not show any variance right so if it's if it's then you will not find a QTL so the QTL profile would just be flat but the CTL profile might show you that there are regions in the genome where the correlation is lost between your phenotype of interest and some other and some other phenotype and so you might find regions of the genome which are not which are not visible using the standard peaks or using standard QTL mapping I hope that it's clear if you have any questions and just ask me I can I can show you in R as well if you if you're interested in that because the the CTL package is just available let me show you in R why not we're here now and we still have 20 minutes right so this was the QTL mapping that we did right so let me go and open up the R window for myself all right so we just do library QTL to load the QTL data set and then we can do library CTL and CTL is the package that I made right so now we are going to use data and we're going to use multi-trade which is the data set that I use for testing so multi-trade is the data set phenotype genotypes that you guys had right so it has 24 phenotypes that's just the same as the one which is on Moodle and if you say pool.geno from multi-trade and then just show a little part 1 to 10 1 to 10 right then you see here that it's exactly the same as before hey you see the same markers they just the individual names are not there but that's because they're already matched inside of this multi-trade object here we can then pool the phenotypes as well which is more or less the same and then you see again that it's the same phenotypes that we saw before so Hydroxypropyl, Hydroxyputanil and so on so if we want to do a QTL scan we can just say plot scan 1 of multi-trade right and then we have to specify which one we want to do so we want to say phenotype column is 1 so just take the first phenotype scan it and you see that you get a profile just as what we did before in the in the standard T testing method hey hey look look at the overview I just got a new follower Jaden Quintana thank you for following me it actually works it actually works they it actually shows you the latest followers yay yay for programming Javascript to interact with the twitch API but let's go back to the CTL or to the QTL mapping so here you see just the same QTL profile as before if we scan a couple of the other ones right then we see that this one only has a single peak so this is phenotype number two like this and then for example we can look at phenotype number three phenotype number four phenotype number five and so on but here at phenotype number five we see something interesting right we do not find a region in the genome which is associated you can see here that the maximum score that we get is around 1.2 right so four we have measured the phenotype we've done a whole big experiment but we find no genetic regulators for this phenotype that's that's that's that's interesting right because we spend a lot of money genotyping these um I'm sorry guys the the mood box crashed when when we went to the to the bunnies I really have to fix that it doesn't crash it just needs a new code to to log into the chat but it when I switch from one scene to the other it it it's just a little bug disappointed yeah I'm disappointed as well I was actually hoping that it would work but at least it still follows the the followers right but here this is so this is really interesting right and and this sometimes happens sometimes you have your phenotype of interest hey you spend a lot of money genotyping like hundreds of individual um you measure the phenotypes and then you do the QTL mapping and you find nothing right so for this phenotype we would just say well oh for phenotype five um which is called um one two three four five so it's for the manual material sulfonyl propyl amount in the arabidopsis plans that we are looking at here we do not find a difference so we wasted a lot of money measuring this material sulfonyl propyl we genotyped individuals and we don't find anything um hey of course we have 24 other phenotypes measured so at least that's good um so we still have something we can publish but normally if you would have only measured this phenotype then now you would have probably wasted around 50 000 euros on your experiment and you would have nothing to be to put in your publication right but fortunately we have measured multiple phenotypes right so because this phenotype five does not have a QTL let's look and see if this one has a CTL right um so we can use the CTL scan function i think that's called CTL scan um and then i'm scan CTL how did i call my function see i think it's capitalized CTL scan yeah so we have CTL scan dot cross which just takes a cross object so we can CTL scan dot cross multi-trade and we just store this in res right so what it will do it will now start doing the um computation for all of the CTLs and this will take some time because it has to do phenotype one versus phenotype two phenotype one versus three one versus four and it has to do that for all of the 24 phenotypes in there fortunately it is relatively optimized so um i spend a lot of time working together with some people in um oakridge national laboratory and they made a new correlation algorithm which is much faster than the one that that i was originally using so they're already finished um when i was doing my phd i had to wait around like 10 minutes for this simple small data set to finish um but now we have our results right so here we have our res and then says that this is a CTL object and we scan 24 phenotypes in total um and now we can just do a plot of this res and we can plot the fifth one right because we're interested in the fifth phenotype um because the fifth phenotype dot did not show a qtl so now it's fingers crossed right and hope that it shows a CTL which one of the other phenotypes so here we then see the output picture that we get and we see that although um wait this is not the fifth one because this has a massive qtl actually um res res number of markers number of phenotypes let me see i think something went wrong and it's actually shown me the wrong three four ah right yeah i use a slightly different scanning method so i do find a qtl for the fifth one um that's a shame and it's like plugging up my all right anyway it doesn't really matter um what i should have done is actually give it the qtl results that i had so let me let me redo that right so we do a scan one of the multi-trade we say phenocall is five and this is our qtl result and now we do a CTL scan of multi-trade pheno dot call is five because we want to do um and then now we can add qtl at qtl is um i think qtl res i didn't mean unused argument let me look at the help it's been been some time ago since i worked on it um ah qtl is true uh yeah use the internal slow qtl mapping method for qtl all right so we can we can do it like this all right anyway like normally you would say that okay we did not find anything but still when we look at this profile we find something interesting right we find that although now using the qtl mapping routine which is in the CTL scan we do find the qtl on chromosome five right so there is something on chromosome five which is more or less determining the average but we see also here that on chromosome four um phenotype number five is actually losing correlation um with the butanil and with the benzo uh benzo wheel hexyl right so on chromosome four there seems to be a regulator which regulates the correlation of this fifth phenotype with two of the other phenotypes and if we would have just done the the qtl scan right we would not have learned that on chromosome four there is a regulator which is involved in the regulation of this phenotype right we would have only learned on chromosome five there is something which is um which is doing uh which is influencing the which is influencing the trait so hey here we see that that using this method you get a additional peak so you get an additional kind of information which is not present in the qtl scan that you did and that is the advantage of using CTL scanning is that it can it can show you where phenotypes lose their correlation or where phenotypes gain correlation um and hey it can even use this to build up a genetic network of which phenotype is controlling which phenotype um and and there's a lot of additional methods in in there to kind of drill down to see exactly what is happening um and you could even just plot the whole thing and then here you see just you see an overview of all of the different markers that we have on the genome hey you see oh this looks a little bit annoying is uh so let's give a little bit more space so we want to give a little bit more space on the bottom definitely more space on the side um and then like this mar so we're just going to set the margin a little bit bigger oh sorry and then we're done so now we can more or less read it right so you see that the first phenotype um has this and so here you see a heat map and you can see here for when we look at the one two three four fifth phenotype here we see that the fifth phenotype seems to be regulated from chromosome four uh it seems to be regulated from chromosome five as well or it loses correlation with other phenotypes so this is just the summary across all of the different phenotypes and then you can look into which phenotypes are causing a correlation loss um for this phenotype at this location so it's it's it's a it's a method which gives you additional information besides standard QTL information um and you also see that hey there's actually three regulators regulating these phenotypes so there's one on chromosome four which is controlling more or less the first couple of phenotypes and there's one on five which is not visible everywhere and there's a massive regulator on chromosome one which is more or less controlling the the deoxy uh methodology i hope that's clear like if they have any questions then feel free to ask me um it's just never do a live demo um because that always goes wrong i actually don't understand why my method finds a QTL at the fifth one while while the standard QTL mapping method doesn't find a QTL there so let's just go um and go through them see if there's any one which i think it's 24 something no that didn't have a significant CTL either yeah this one is very clear that there's a QTL and a CTL at the same place here as well so we don't find a new genetic locus here also not and so the idea is is that it gives you additional information but here you see that just a single QTL scan teaches you that phenotype number 20 is controlled from chromosome one but it's not only controlled from chromosome one it also loses the correlation with the uh isohumnectin deoxy but that's the idea between CTL mapping is that you don't just get a genetic locus which which is controlling your trait of interest but also shows you how the correlation of this trait is lost with other phenotypes and then the fifth one is a quite nice example because you would have never said something on chromosome four is driving this phenotype just looking at the QTL profile you would have only looked at chromosome five but looking at the CTL profile you would now get the idea there it might be worth my time also investigating chromosome four and see why this phenotype is losing correlation on chromosome four uh combined to the other one uh combined to the other phenotypes all right so that's just what I wanted to show you guys um software is available it's called library CTL um and hey in case that you ever do a QTL scan and hey you've measured multiple phenotypes and one of your phenotypes just doesn't show a QTL for some reason um and then it's definitely worth it just trying the CTL scan method um to figure out um if you can find genetic regulators not that regulate your phenotype directly but which break your link between genotype or which which control the the correlation between your phenotype of interest and some of the other phenotypes that you've measured all right at least I'm going to stop the recording for the people watching on Moodle so um see you guys next week on Moodle and I will upload the the assignments as soon as possible for the primer design lecture as well all right