compare_pubmed_results_geographically

Xpath Expression

/PubmedArticleSet/PubmedArticle/MedlineCitation/Article/AuthorList/Author/AffiliationInfo/Affiliation

Script

/*
Inputs: 	country_code (1)
	country_name (1)
	affiliations (1)
				
Outputs:	country_list (1)
	count_list   (1)
*/

	country_list = new ArrayList(); //list of all countries in 3 letter codes
	count_list = new ArrayList(); //list of counts, corresponding to countries
	matched_countries = new ArrayList(); //countries are matched for a single affiliation
	string_index = new ArrayList(); //keeps indexes of matched countries for each affiliation
	string_length = new ArrayList(); //keeps the string length of matched countries for each affiliation 
	country_matches = new ArrayList(); //matched countries after index filter (which picks the last) for each affiliation 	
	int index1,index2;
	int length1,length2;
	
	//key->3 letter country code, value->affiliation count
	map = new Hashtable();
		
	//creates unified country list with default 0s 
	for(int i=0; i=0){
				//add the corresponding 3-letter country code to matched_countries list of that affiliation
				matched_countries.add(country_code.get(j));
				//add the index of the country_name to string_index list of that affiliation
				string_index.add(si);
				//add the length of the matched country_name to string_length list of that affiliation
				string_length.add(country_name.get(j).length());	
	
			}
		}
			
		try{
			//get the first element of matched_countries (Exception if the list is empty)
			country_matches.add(matched_countries.get(0));
			//get the index of the first country_name that is appearing in the affiliation
			index1 = string_index.get(0);
			//set country the value which has the largest index (the country_name match that appears near the end)
			for(int k=1;kindex1){
					country_matches.clear();
					country_matches.add(matched_countries.get(k));
				}
				else if(index2==index1){
					country_matches.add(matched_countries.get(k));
				}
				index1=index2;
			}
			//get the first element of country_matches (Exception if the list is empty)
			country = country_matches.get(0);
			//get the length of the first country_name that is appearing in the affiliation
			length1 = string_length.get(0);
			//set country the value which is the longest match (US vs USSR)
			for(int k=1;klength1){
					country = country_matches.get(k);
				}
				length1=length2;
			}

			//increase the count for the country which is matched for the affiliation
			temp1=map.get(country)+1;
			map.put(country,temp1);
		}
		catch (Exception e){
			//The affiliation is not matched with any of the countries
		}
		//Reset the arraylists for the next iteration of affiliation
		matched_countries.clear();
		string_index.clear();
		string_length.clear();
		country_matches.clear();
	}
		
	//get the counts for each country and copy them to the output arraylist
	for(int i=0; i

Script

/*
	Inputs: 	meta_country_list      (1)
				meta_count_list        (1)
				former_countries_list  (1)
				current_countries_list (1)
				
	Outputs:	country_list (1)
				count_list   (1)
*/

	country_list = new ArrayList();
	count_list = new ArrayList();
	successor_countries = new ArrayList();
	
	//creates hashtable with country codes as keys and counts as values 
	country_count = new Hashtable();
	
	for(int i=0; i

Script

#Inputs:  countries (String vector)
#         counts    (Integer vector)
#
#Outputs: output (String)

library(rworldmap)
affiliation<-as.data.frame(matrix(nrow=length(countries), ncol=2));
affiliation[,1]<-data.matrix(countries);
affiliation[,2]<-data.matrix(counts);
total<-sum(counts);

sPDF<-joinCountryData2Map(affiliation, joinCode="ISO3", nameJoinColumn="V1");

output<-"output.png";
png(output);
par(mai=c(0,0,0.2,0),xaxs="i",yaxs="i");

mapCountryData(sPDF, nameColumnToPlot="V2", catMethod = exp(seq(from=0, to=log(max(affiliation[,2])), length.out=100)), colourPalette = "heat", addLegend = TRUE, borderCol ='black', mapTitle ='Publications on mass spectrometry in PubMed');
dev.off();

R Server