go_kegg_pathways

go_kegg_pathways.r

library(dplyr)
library(org.Hs.eg.db)

# get KEGG pathways

xx <- as.list(org.Hs.egPATH2EG)
keggpathways <- lapply(names(xx), function(x) {
	tibble(
		symbol=select(org.Hs.eg.db, xx[[x]], c("SYMBOL"), "ENTREZID")$SYMBOL,
		pathway=x
	)
}) %>% bind_rows()



# get GO pathways
# this will take a bit of time - there are 18000 GO terms or something

yy <- as.list(org.Hs.egGO2EG)
gopathways <- lapply(names(yy), function(x) {
	tibble(
		symbol=select(org.Hs.eg.db, yy[[x]], c("SYMBOL"), "ENTREZID")$SYMBOL,
		evidence=names(yy[[x]]),
		pathway=x
	)
}) %>% bind_rows()


# create some example data
# two protein pairs that are in the same pathways, one which isn't

pqtl_data <- data.frame(
	pqtl1 = c("A2M", "CYP3A4", "abc"),
	pqtl2 = c("SERPINC1", "TYMP", "cde")
)

# function to get overlaps
get_joint_pathways <- function(pqtl_data, pathwaydata)
{
	d1 <- inner_join(pqtl_data, pathwaydata, by=c("pqtl1"="symbol"))
	d1 <- inner_join(d1, pathwaydata, by=c("pqtl2"="symbol"))
	d2 <- d1 %>% 
		mutate(samepathway = pathway.x==pathway.y) %>%
		subset(., samepathway) %>%
		subset(!duplicated(paste(pqtl1, pqtl2)))
	return(nrow(d2))
}

get_joint_pathways(pqtl_data, keggpathways)

# function to do permutations
# might be slow
get_joint_pathways <- function(pqtl_data, pathwaydata, nperm)
{
	o <- rep(NA, nperm)
	for(i in 1:nperm)
	{
		pqtl_data$pqtl1 <- sample(pqtl_data$pqtl1)
		pqtl_data$pqtl2 <- sample(pqtl_data$pqtl2)
		d1 <- inner_join(pqtl_data, pathwaydata, by=c("pqtl1"="symbol"))
		d1 <- inner_join(d1, pathwaydata, by=c("pqtl2"="symbol"))
		d2 <- d1 %>% 
			mutate(samepathway = pathway.x==pathway.y) %>%
			subset(., samepathway) %>%
			subset(!duplicated(paste(pqtl1, pqtl2)))
		o[i] <- nrow(d2)
	}
	return(o)
}