@@DannyArends Thank you for your effort and for your generosity I'm a fun of you especially since I have started this bioinformatics adventure. Hopefully one day I can visit you in your lab. Thanks once again .. I'm following you! Best Regards,
Thank you for the nice lectures : ) ! By any chance, do you plan (or maybe you already have and I overlooked) to make a video like a step-by-step tutorial on QTL analysis in R ?
Not yet, currently working on getting an "Your own AI voice from scratch" lecture put together. However, a QTL mapping from scratch lecture is definitely going to be next.
Awesome video, is very useful for people who is starting to learn about this topic (like me), could you provide me the data set! I would like to practice. Thanks for your time!!
You mean the blood pressure data set in mice ? This one is available as the data set hyper in R/qtl, in R issue the following commands to get the data set: # Install the qtl package install.packages("qtl") # Load the qtl package and the hyper data set library("qtl") data(hyper) # Get the genotypes and phenotype matrices geno
Interesting video. I got 82 controls and 116 cases. I encoded my genotypes as follows: 0/0 = 0; 0/1 or 1|0 = 1; 1/1 = 2. I split cases and controls in two matrices. For each row, I counted the number of zeros and divided the number by the number of cases for the cases or the number of controls for the controls. I then subtracted the two from each other to get the difference. Should I rather perform a fisher-exact test instead of just subtracting the two values from each other? I am also currently trying to apply the knowledge you share, but I somehow stugle with how I could do that. Help would be appreciated.
In this case logistic regression is probably going to be what you want. Create a phenotype for every individual which is it's case/control status, then use regression using the glm() function and set your link function accordingly within the regression. Something along the lines of: glm(status ~ sex + covX + marker, family=binomial(link='logit')) This will give you most power to detect if any genetic marker influence your phenotype, and it's going to allow you to compensate for other covariates like sex, age, etc See an example here: www.r-bloggers.com/2015/09/how-to-perform-a-logistic-regression-in-r/
Just FANTASTIC !! Bravo!
Thanks, glad you like the lectures :)
@@DannyArends Thank you for your effort and for your generosity I'm a fun of you especially since I have started this bioinformatics adventure. Hopefully one day I can visit you in your lab. Thanks once again .. I'm following you! Best Regards,
The best presentation on qtl ever.
Thanks 👍
Wow, thank you so much
You're welcome, hope it was useful
Thank you Prof!
You are welcome!
Gorgeous! Thank you!
Thank you for the nice lectures : ) ! By any chance, do you plan (or maybe you already have and I overlooked) to make a video like a step-by-step tutorial on QTL analysis in R ?
Thanks for the suggestion, it's definitely something I can consider/see myself doing. QTL mapping from scratch.
Did this happened@@DannyArends ?
Not yet, currently working on getting an "Your own AI voice from scratch" lecture put together. However, a QTL mapping from scratch lecture is definitely going to be next.
Thank you prof
You are welcome
Awesome video, is very useful for people who is starting to learn about this topic (like me), could you provide me the data set! I would like to practice. Thanks for your time!!
You mean the blood pressure data set in mice ?
This one is available as the data set hyper in R/qtl, in R issue the following commands to get the data set:
# Install the qtl package
install.packages("qtl")
# Load the qtl package and the hyper data set
library("qtl")
data(hyper)
# Get the genotypes and phenotype matrices
geno
Interesting video. I got 82 controls and 116 cases. I encoded my genotypes as follows: 0/0 = 0; 0/1 or 1|0 = 1; 1/1 = 2. I split cases and controls in two matrices. For each row, I counted the number of zeros and divided the number by the number of cases for the cases or the number of controls for the controls. I then subtracted the two from each other to get the difference. Should I rather perform a fisher-exact test instead of just subtracting the two values from each other? I am also currently trying to apply the knowledge you share, but I somehow stugle with how I could do that. Help would be appreciated.
In this case logistic regression is probably going to be what you want. Create a phenotype for every individual which is it's case/control status, then use regression using the glm() function and set your link function accordingly within the regression.
Something along the lines of:
glm(status ~ sex + covX + marker, family=binomial(link='logit'))
This will give you most power to detect if any genetic marker influence your phenotype, and it's going to allow you to compensate for other covariates like sex, age, etc
See an example here: www.r-bloggers.com/2015/09/how-to-perform-a-logistic-regression-in-r/
@@DannyArends thank you for your quick response. I used PLINK for an association analysis.