{"id":1957,"date":"2015-04-22T16:28:36","date_gmt":"2015-04-22T23:28:36","guid":{"rendered":"http:\/\/lukemiller.org\/?p=1957"},"modified":"2015-04-22T17:36:44","modified_gmt":"2015-04-23T00:36:44","slug":"a-plot-of-co-authorships-in-my-little-corner-of-science","status":"publish","type":"post","link":"https:\/\/lukemiller.org\/index.php\/2015\/04\/a-plot-of-co-authorships-in-my-little-corner-of-science\/","title":{"rendered":"A plot of co-authorships in my little corner of science"},"content":{"rendered":"

\"author\n<\/p>\n

 
\nHere’s a mostly useless visualization of the collection of journal articles that sits in my reference database in Endnote. I deal mostly in marine biology, physiology, biomechanics, and climate change papers, with a few molecular\/genetics papers thrown in here and there. The database has 3325 entries, 2 of which have ambiguous publication years and aren’t represented above. This is by no means an exhaustive survey of the literature in my field, it’s just an exhaustive survey of the literature on my computer.<\/p>\n

To make this figure, I first had Endnote export the database to a text file using an output style that simply listed the publication year and the authors, each separated by commas:<\/p>\n

2009, L. W. Aarssen, C. J. Lortie, A. E. Budden, J. Koricheva, R. Leimu, T. Tregenza\r\n1956, D. P. Abbott\r\n1968, D. P. Abbott, D. Epel, J. H. Phillips, I. A. Abbott\r\n1983, A. H. Abdel-Rehim\r\n1983, A. H. Abdel-Rehim\r\n1988, A. H. Abdel-Rehim\r\n2003, M. A. Abdelrhman\r\n2007, M. A. Abdelrhman\r\n1997, A. Abelson, M. Denny\r\n.\r\n.\r\n.\r\n<\/pre>\n
\"In<\/a>
In Endnote, the bibliography output style is set to show only the Year and Author, using a single Generic style that will get applied to all reference types.<\/figcaption><\/figure>\n
The R script below then parses that text file to pull out the year of each publication and the number of authors for each publication. The number of authors ranged between 1 and 121 (that 121 author paper is purposely cut off in the figure). I then tallied the number of papers that fell into each combination of publication year and number of co-authors. That count is then translated into one of the color values represented in the color scalebar. The colorbar is a half-hearted attempt to map the range of count values (1 to 69 in this case) onto a range of colors that is perceived as fairly linear by the human eye, based on the recommendations at http:\/\/earthobservatory.nasa.gov\/blogs\/elegantfigures\/2013\/08\/05\/subtleties-of-color-part-1-of-6\/<\/a> and ColorBrewer2.org<\/a>. The function for plotting the colorbar was derived from this posting.<\/a><\/p>\n
Unsurprisingly, to me at least, 1, 2, and 3 author papers are the most numerous, and I clearly have pulled more papers from the late 90’s and 2000’s than I have from the earlier literature. Papers with more than 9 or 10 co-authors are fairly rare in my collection, with most of those levels being represented by just one or a few papers. <\/p>\n
The R script author_year_plot.R<\/code> and the associated text data file authors_list_20150422.txt<\/code> are in my GitHub repository<\/a> if you’re bored enough to want to try your hand at recreating the figure. <\/p>\n
 <\/p>\n
\r\n# author_year_plot.R\r\n# \r\n# Author: Luke Miller 2015-04-22\r\n###############################################################################\r\n\r\n\r\n###############################################################################\r\n# Export a text file from Endnote that only lists Year and Authors, all \r\n# separated by commas. To do this, create an Output Style\r\n# that lists the year followed by a comma and then each author separated by\r\n# a comma. Select all references, then go to File>Export. In the window that\r\n# opens, you'll see a menu for output style, choose your author-only version \r\n# there and save the output file as text file. \r\nf1 = 'authors_list_20150422.txt'\r\n#\r\n## Scan input file, divide each line into a separate entry in a character vector\r\nauthors = scan(file = f1, what = character(), sep = '\\n') \r\n#\r\nyr = character()\r\n# Extract year from each record. \r\nfor (i in 1:length(authors)){\r\n\tyr[i] = substr(authors[i],regexpr('[1-2]',authors[i])[[1]],\r\n\t\t\tregexpr(',',authors[i])[[1]] - 1)\r\n}\r\nyr = as.numeric(yr) # Convert to numbers\r\n# Entries with missing or ambiguous years (anything with multiple years listed\r\n# like 1997-2013) will end up as NA's in the yr vector, and will generate a \r\n# warning.\r\n\r\ncnt = numeric(length(yr)) # Create empty vector\r\n# To count the number of authors on a paper, simply count the number of \r\n# commas in each line of the authors vector. There is always one comma after \r\n# the year, denoting at least one author, and every additional comma means there\r\n# is another author. \r\nfor (i in 1:length(authors)){\r\n\tcnt[i] = length(gregexpr(',',authors[i])[[1]])\r\n}\r\n# Pick out rows that don't have a useful year value\r\nbad.entries = which(is.na(yr))\r\n# Remove the offending rows from the yr and cnt vectors\r\nyr = yr[-(bad.entries)]\r\ncnt = cnt[-(bad.entries)]\r\n# Make a data frame out of the yr and cnt vectors\r\ndf = data.frame(Year = yr, Count = cnt)\r\n\r\n# Make a new dataframe that holds each combination of Year and Count\r\nnewdf = expand.grid(Years = unique(yr), Count = unique(cnt))\r\n# Make a new column to hold a tally of the number of papers for each Year and\r\n# author Count combination. \r\nnewdf$TotalPapers = NA\r\n\r\n# Go through the combinations of years and counts to tally the number of papers\r\n# that match that combo in the 'df' dataframe\r\nfor (i in 1:nrow(newdf)){\r\n\t# Put the tally of number of papers matching each Year & Count combo in the\r\n\t# TotalPapers column\r\n\tnewdf$TotalPapers[i] = nrow(df[df$Year == newdf$Year[i] & \r\n\t\t\t\t\t\t\tdf$Count == newdf$Count[i],])\r\n}\r\n\r\n# Drop any combinations where the TotalPapers was 0\r\nnewdf = newdf[-(which(newdf$TotalPapers == 0)),]\r\n\r\n#########################################################\r\n#########################################################\r\n# Create a function to plot a color scale bar on the existing plot using the\r\n# vector of colors that will be generated later by the colorRampPalette function\r\ncolor.bar <- function(lut, min, max=-min, nticks=11, \r\n\t\tx1 = 1, x2 = 2, y1 = 1, y2 = 2, \r\n\t\tticks=seq(min,max, length=nticks), round = TRUE, title = '',\r\n\t\tcex.title = 1, text.col = 'black', horiz = FALSE){\r\n\t# lut = a vector of color values, in hex format\r\n\t# min = minimum value represented by the first color\r\n\t# max = maximum value represented by the last color\r\n\t# nticks = number of tick marks on the colorbar\r\n\t# x1 = location of left edge of colorbar, in plot's x-units\r\n\t# x2 = location of right edge of colorbar, in plot's x-units\r\n\t# y1 = location of bottom edge of color bar, in plot's y-units\r\n\t# y2 = location of top edge of color bar, in plot's y-units\r\n\t# ticks = a sequence of tick mark value to be added to colorbar\r\n\t# round = TRUE or FALSE, round off tick values to 0 decimal place.\r\n\t# title = Title for colorbar\r\n\t# cex.title = size for title\r\n\t# text.col = color of tick marks, title, and border of colorbar\r\n\t# horiz = TRUE or FALSE, lay out color bar vertically or horizontally\r\n\t\r\n\t# Calculate a scaling factor based on the number of entries in the \r\n\t# look-up-table and the absolute distance between y2 and y1 on the plot\r\n\tif (horiz == FALSE){\r\n\t\tscale = (length(lut)-1)\/(y2-y1)\t\r\n\t} else if (horiz == TRUE){\r\n\t\t# For horizontal bars, use the distance between x2 and x1 instead\r\n\t\tscale = (length(lut)-1)\/(x2-x1)\r\n\t}\r\n\t# Round off the tick marks if desired\r\n\tif (round) { ticks = round(ticks,0) }\r\n\t# Draw little thin rectangles for each color in the look up table. The\r\n\t# rectangles will span the distance between x1 and x2 on the plot's \r\n\t# coordinates, and have a y-axis height scaled to fit all of the colors\r\n\t# between y1 and y2 on the plot's coordinates. Each color will only be a\r\n\t# small fraction of that overall height, using the scale factor. For a \r\n\t# horizontal-oriented bar the thin rectangles will run between y1 and y2,\r\n\t# scaled to fit all of the colors between x1 and x2. \r\n\tfor (i in 1:(length(lut)-1)) {\r\n\t\tif (horiz == FALSE) {\r\n\t\t\t# Calculate myy, the lower y-location of a rectangle\r\n\t\t\tmyy = (i-1)\/scale + y1\r\n\t\t\t# Calculate the upper y value as y+(1\/scale), and draw the rectangle\r\n\t\t\trect(x1,myy,x2,myy+(1\/scale), col=lut[i], border=NA)\r\n\t\t} else if (horiz == TRUE) {\r\n\t\t\t# Calculate x, the left x-location of a rectangle\r\n\t\t\tmyx = (i-1)\/scale + x1\r\n\t\t\t# Calculate the right x value as x+(1\/scale), and draw the rectangle\r\n\t\t\trect(myx,y1,myx+(1\/scale),y2, col=lut[i], border=NA)\r\n\t\t}\r\n\t}\r\n\t# Draw a border around the color bar\r\n\trect(x1,y1,x2,y2, col = NULL, border = text.col)\r\n\t# Draw tick marks and tick labels\r\n\tfor (i in 1:length(ticks)){\r\n\t\tif (horiz == FALSE) {\r\n\t\t\tmyy = (ticks[i]-1)\/scale + y1\r\n\t\t\t# This is an attempt to set the tick mark and labels just off to the \r\n\t\t\t# right side of the color bar without having them take up too much \r\n\t\t\t# of the plot area. The x locations are calculated as x2 plus a \r\n\t\t\t# fraction of the width of the rectangle.\r\n\t\t\tmyx2 = x2 + ((x2-x1)*0.1)\r\n\t\t\tmyx3 = x2 + ((x2-x1)*0.13)\r\n\t\t\t# Draw little tick marks\r\n\t\t\tlines(x = c(x2,myx2), y = c(myy,myy), col = text.col)\r\n\t\t\t# Draw tick labels\r\n\t\t\ttext(x = myx3, y = myy, labels = ticks[i], adj = c(0,0.3), \r\n\t\t\t\t\tcol = text.col)\r\n\t\t} else if (horiz == TRUE) {\r\n\t\t\t# For a horizontal scale bar\r\n\t\t\tmyx = (ticks[i]-1)\/scale + x1\r\n\t\t\t\r\n\t\t\t# This is an attempt to set the tick mark and labels just below the \r\n\t\t\t# bottom of the color bar without having them take up too much of\r\n\t\t\t# the plot area. The y locations are calculated as y1 minus a \r\n\t\t\t# fraction of the height of the rectangle\r\n\t\t\tmyy2 = y1 - ((y2-y1)*0.1)\r\n\t\t\tmyy3 = y1 - ((y2-y1)*0.13)\r\n\t\t\t# Draw little tick marks\r\n\t\t\tlines(x = c(myx,myx), y = c(y1,myy2), col = text.col)\r\n\t\t\t# Draw tick labels\r\n\t\t\ttext(x = myx, y = myy3, labels = ticks[i], adj = c(0.5,1), \r\n\t\t\t\t\tcol = text.col)\r\n\t\t}\r\n\t}\r\n\t# Draw a title for the color bar\r\n\ttext(x = ((x1+x2)\/2), y = y2, labels = title, adj = c(0.5,-0.35),\r\n\t\t\tcex = cex.title, col = text.col)\r\n}\r\n####################################################\r\n####################################################\r\n# Define a color ramp function from white to blue\r\n\r\n# From ColorBrewer 9-class Blues (single-hue). ColorBrewer recommends the \r\n# following set of 9 color values, expressed in hex format. I reverse them so\r\n# that the highest value will be the lightest color. \r\ncolfun = colorRampPalette(rev(c(\"#f7fbff\",\"#deebf7\",\"#c6dbef\",\"#9ecae1\",\r\n\t\t\t\t\t\t\"#6baed6\",\"#4292c6\",\"#2171b5\",\"#08519c\",\"#08306b\")),\r\n\t\tspace = 'Lab')\r\n\r\n# Define a set of colors from blue to white using that function, covering the\r\n# entire range of possible values for newdf$TotalPapers\r\ncols = colfun(max(newdf$TotalPapers))\r\n# Assign a color to each entry in the newdf data frame based on its TotalPapers\r\n# value. \r\nnewdf$col = \"\"\r\nfor (i in 1:nrow(newdf)){\r\n\tnewdf$col[i] = cols[newdf$TotalPapers[i]]\r\n}\r\n\r\n##############################\r\n# Create an output file in svg format\r\nsvg(filename = \"author-year-count.svg\", width = 9, height = 4.8)\r\npar(mar =c(5,6,1,2)) # Change the figure margins slightly\r\nplot(Count~Years, data = newdf, type = 'n', \r\n\t\tylim = c(0,45), las = 1,\r\n\t\tcex.lab = 1.6,\r\n\t\tcex.axis = 1.3,\r\n\t\tylab = 'Number of coauthors',\r\n\t\txlab = 'Publication Year',\r\n\t\tyaxt = 'n')\r\n# Color the background of the plot using a rectangle, and determine its \r\n# dimensions on the fly by calling the par()$usr function to get the coordinates\r\n# of the plot edges.\r\nrect(par()$usr[1],par()$usr[3],par()$usr[2],par()$usr[4], col = \"#BBBBBB\")\r\n# Draw some grid lines at useful locations\r\nabline(h = c(1,2,3,4,5,10,15,20,25,30,35,40), col = \"#CCCCCC\")\r\nabline(v = seq(1875,2015, by = 5), col = \"#CCCCCC\")\r\n# Redraw the plot's bounding box to cover where the horizontal lines overwrite\r\n# it. \r\nbox()\r\n# Redraw the point data over the newly drawn background and horizontal lines\r\npoints(Count~Years, data = newdf, col = newdf$col, pch = 20, cex = 0.9)\r\n# Call the color.bar function created earlier to create a color scale.\r\ncolor.bar(lut = cols, nticks = 8, horiz = TRUE,\r\n\t\tmin = 1, max = max(newdf$TotalPapers),\r\n\t\tx1 = 1880, x2 = 1920, y1 = 42, y2 = 44, \r\n\t\ttitle = 'Number of papers', cex.title = 1.1, text.col = 'black')\r\n# Draw the y-axis labels at the appropriate spots\r\naxis(2, at = c(1,2,3,4,5,10,15,20,25,30,35,40), \r\n\t\tlabels = c('1','','3','','5','10','15','20','25','30','35','40'), \r\n\t\tlas = 1, cex.axis = 1.1)\r\ndev.off()\r\n\r\n\r\n\r\n\r\n\r\n<\/pre>\n","protected":false},"excerpt":{"rendered":"
  Here’s a mostly useless visualization of the collection of journal articles that sits in my reference database in Endnote. I deal mostly in marine biology, physiology, biomechanics, and climate change papers, with a few molecular\/genetics papers thrown in here and there. The database has 3325 entries, 2 of which have ambiguous publication years and […]<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[4,218],"tags":[210,211,17,58],"class_list":["post-1957","post","type-post","status-publish","format-standard","hentry","category-journal","category-r-project","tag-color-palette","tag-color-ramp","tag-plotting","tag-r-project"],"_links":{"self":[{"href":"https:\/\/lukemiller.org\/index.php\/wp-json\/wp\/v2\/posts\/1957","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/lukemiller.org\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/lukemiller.org\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/lukemiller.org\/index.php\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/lukemiller.org\/index.php\/wp-json\/wp\/v2\/comments?post=1957"}],"version-history":[{"count":16,"href":"https:\/\/lukemiller.org\/index.php\/wp-json\/wp\/v2\/posts\/1957\/revisions"}],"predecessor-version":[{"id":1974,"href":"https:\/\/lukemiller.org\/index.php\/wp-json\/wp\/v2\/posts\/1957\/revisions\/1974"}],"wp:attachment":[{"href":"https:\/\/lukemiller.org\/index.php\/wp-json\/wp\/v2\/media?parent=1957"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/lukemiller.org\/index.php\/wp-json\/wp\/v2\/categories?post=1957"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/lukemiller.org\/index.php\/wp-json\/wp\/v2\/tags?post=1957"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}