A couple of days ago, I was looking for a way to create a citation graph from list of DOIs using R.

It turned out that, although packages exists to get basic citation information from Google Scholar (with scholar), Scopus (rscopus) or CrossRef (rcrossref), informations about citing article is pretty much closed. The only site that allowed it was PubMed, with the drawback that it only analyse its own database. Here is how I did it.

First, we start with a list PubMed ids, for which we would like to have the citation graph. In this exemple, these are the id’s from my own papers. We also store a the id’s as a single string (to be re-used latter).

PMIDs <- c(27352932,
           26905656,
           26476447,
           26287479,
           25657812,
           25614065,
           24744766,
           24515834,
           24143806,
           24107223,
           23875292,
           23286457,
           22558767,
           21771915,
           20453027)

PMIDs_ch <- paste(PMIDs, collapse=",")

Then, for each PMID, we ask PubMed to return the prides of the citing article. This is simply done by building a custom url, thanks to the NCBI API’s.

# We build the URL to retrieve the citing PMIDs
path <- paste("https://www.ncbi.nlm.nih.gov/pubmed?linkname=pubmed_pubmed_citedin&from_uid=",
                PMID,
                "&report=uilist&format=text&dispmax=200", sep="")
f <- file(path)
data <- readLines(f, warn = F)
data <- gsub("&lt;", "<", data)
data <- gsub("&gt;", ">", data)
citing <-strsplit(xmlToList(xmlParse(data, asText = T))[1], "\n")[[1]]
close(f)

Ones we have the citing PMIDs as a list, we create a new data frame containing with the source PMID, the citing PMIDs.

# Then we create a table containing all these PMIDs
if(length(citing) < 200){
for(pm in citing){
  citegraph <- rbind(citegraph, 
	  data.frame(PMID=as.numeric(PMID), 
							  citing=as.numeric(pm)))
}
}

Finally, we use visNetwork for the visualisation of the citation graph. We use a different colour for the source PMIDs and the citing PMIDs.

# We use visNetwork for the vizualisation of the network
# For that we need a nodes and edges tables. 

nodes <- data.frame(id=unique(c(citegraph$PMID, citegraph$citing)))
for(i in 1:nrow(nodes)){
  nodes$color[i] <- "lightblue"
  if(grepl(nodes$id[i], PMIDs_ch)){
    nodes$color[i] <- "red"
  }
}

edges <- data.frame(from = citegraph[,2], to = citegraph[,1])
visNetwork(nodes, edges, width = "100%")

The complete code I used is available here