Buck9195/Gladstone-Bioinformatics-Workshop
1
0
Fork 0
mirror of https://github.com/gladstone-institutes/Bioinformatics-Workshops.git synced 2025-11-30 09:45:43 -08:00
Code Issues Projects Releases Packages Wiki Activity Actions

Add files via upload

Browse source
This commit is contained in:
MichelaTr 2023-11-25 17:22:51 -08:00 • committed by GitHub
parent 00aaa4e8be
commit b7962963e9
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 6 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

9
GWAS_PRS/scripts/step06_best_PRS_visualization.R
View file

@ -13,14 +13,14 @@ phenotype <- read.csv("EUR_input_for_PRS/EUR_BMI_phenofile.csv", header=T, check
# To make things simple, we will add the appropriate headers # To make things simple, we will add the appropriate headers
# (1:6 because there are 6 PCs) # (1:6 because there are 6 PCs)
#colnames(pcs) <- c("FID", "IID", paste0("PC",1:6)) #colnames(pcs) <- c("FID", "IID", paste0("PC",1:6))
# Read in the covariates (here, it is sex) # Read in the covariates (here, it is sex and age)
covariate <- read.csv("EUR_input_for_PRS/EUR_covariates.csv", header=T,check.names=F, comment.char="") covariate <- read.csv("EUR_input_for_PRS/EUR_covariates.csv", header=T,check.names=F, comment.char="")
# Now merge the files # Now merge the files
pheno <- merge(phenotype, covariate, by=c("IID")) pheno <- merge(phenotype, covariate, by=c("IID"))
#pheno <- merge(merge(phenotype, covariate, by=c("FID", "IID")), pcs, by=c("FID","IID")) #pheno <- merge(merge(phenotype, covariate, by=c("FID", "IID")), pcs, by=c("FID","IID"))
# We can then calculate the null model (model with PRS) using a linear regression # We can then calculate the null model (model with PRS) using a linear regression
# (as height is quantitative) # (as bmi is quantitative)
null.model <- lm(BMI~., data=pheno[,-1]) null.model <- lm(BMI~., data=pheno[,-1])
# And the R2 of the null model is # And the R2 of the null model is
null.r2 <- summary(null.model)$r.squared null.r2 <- summary(null.model)$r.squared
@ -32,7 +32,7 @@ for(i in p.threshold){
# We only want the FID, IID and PRS from the PRS file, therefore we only select the # We only want the FID, IID and PRS from the PRS file, therefore we only select the
# relevant columns # relevant columns
pheno.prs <- merge(pheno, prs[,c("IID", "SCORE1_AVG")], by= "IID") pheno.prs <- merge(pheno, prs[,c("IID", "SCORE1_AVG")], by= "IID")
# Now perform a linear regression on Height with PRS and the covariates # Now perform a linear regression on BMI with PRS and the covariates
# ignoring the FID and IID from our model # ignoring the FID and IID from our model
model <- lm(BMI~., data=pheno.prs[,-1]) model <- lm(BMI~., data=pheno.prs[,-1])
# model R2 is obtained as # model R2 is obtained as
@ -65,5 +65,8 @@ plot(x=dat$SCORE1_AVG, y=dat$BMI, col="white",
xlab="Polygenic Score", ylab="BMI") xlab="Polygenic Score", ylab="BMI")
with(subset(dat, sex=="Male"), points(x=SCORE1_AVG, y=BMI, col="red", pch=19)) with(subset(dat, sex=="Male"), points(x=SCORE1_AVG, y=BMI, col="red", pch=19))
with(subset(dat, sex=="Female"), points(x=SCORE1_AVG, y=BMI, col="blue", pch=19)) with(subset(dat, sex=="Female"), points(x=SCORE1_AVG, y=BMI, col="blue", pch=19))
plot(density(dat$SCORE1_AVG[dat$sex =="Male"]), col="red", lwd=3)
lines(density(dat$SCORE1_AVG[dat$sex =="Female"]), col="blue", lwd=3)
dev.off() dev.off()
#################################################### ####################################################