mesh

mesh.R

query= 17554300
#run the search
search <- EUtilsSummary(query, type="esearch",db = "pubmed",retmax=30000)
QueryCount(search) #how many hits
summary(search) #number of hits and search terms used
records <- EUtilsGet(search) #get the results from search

PMID(records)
Author(records)[[1]]
a <- Mesh(records)[[1]]
class(a)



[[1]]


mesh <- lapply(pmid[1:10], function(x)
{
	print(x)
	search <- EUtilsSummary(x, type="esearch",db = "pubmed",retmax=30000)
	records <- EUtilsGet(search) #get the results from search
	Mesh(records)[[1]]
})


library(plyr)
library(RISmed)


# For every PMID for which there is only ONE phenotype, download the mesh terms

...



# For every PMID for which there are multiple phenotypes, download the mesh terms
gwas_catalog$code <- paste(gwas_catalog$PubmedID, gwas_catalog$Phenotype)
g <- subset(gwas_catalog, !duplicated(code))

tab <- table(g$PubmedID)
n <- names(tab[tab>1])
head(sort(tab, decreasing=T))

g2 <- subset(g, PubmedID %in% n, select=c(Phenotype, PubmedID))

mesh <- lapply(unique(g2$PubmedID), function(x)
{
	print(x)
	search <- EUtilsSummary(x, type="esearch",db = "pubmed",retmax=30000)
	records <- EUtilsGet(search) #get the results from search
	Mesh(records)[[1]]
})


names(mesh) <- unique(g2$PubmedID)
mesh2 <- mesh[sapply(mesh, is.data.frame)]


for(i in 1:length(mesh2))
{
	mesh2[[i]]$PubmedID <- names(mesh2)[i]
}

x <- rbind.fill(mesh2)

x1 <- ddply(x, .(PubmedID), function(x)
{
	x <- mutate(x)
	y <- paste(x$Heading, collapse="\t")
	x <- x[1,-1]
	x$Headflat <- y
	x
})


tab <- table(x$Heading)
n <- names(tab[tab > 10])
x1 <- subset(x, ! Heading %in% n)

x2 <- ddply


x2 <- merge(x1, g2, by="PubmedID")