###############################################################
#Project
#josh dornfeld
###############################################################

###############################################################
# 	Identifying	differentially	expressed	genes	between	males	and
# females with asthma
###############################################################


library(limma)
library(class)
library(GEOquery)
rm(list = ls())

#downloading data
GSE43696 = getGEO("GSE43696")
GSE43696

#Pulling out gene expression data and pheno type data
GSE43696.expr = exprs(GSE43696[[1]])
boxplot(GSE43696.expr[, 0:50], main = "processed data")
GSE43696.p = pData(GSE43696[[1]])

#pulling out gender column
gender = as.character(GSE43696.p$characteristics_ch1.2)
table(gender)


levels(gender) = c("gender: female", "gender: male")
design = model.matrix(~0+gender)
colnames(design)
colnames(design) = c("Female", "Male")
fit = lmFit(GSE43696.expr, design)
contrast.matrix <- makeContrasts(Female - Male,levels=design)
fit = contrasts.fit(fit, contrast.matrix)
fit = eBayes(fit)
tt.05 = topTable(fit,sort.by = "p", p.value = 0.2, number = nrow(GSE43696.expr))