Proofing

Author: PierreMasselot

1b_CompositionSummary.R

@@ -21,8 +21,8 @@ tot_spec <- do.call(rbind, dlist_spec)
 #Useful objects for later
 spec_inds <- grep("PM25", colnames(tot_spec))
 spec_names <- c("SO4", "NH4", "NO3", "BC", "OC", "SS", "DUST")
-spec_labs <- c(expression(SO[4]^{"2-"}), expression(NH[4]^{"+"}), expression(NO[3]^{"-"}), 
-  "BC", "OC", "SS", "DUST")
+spec_labs <- c(expression(SO[4]^{"2-"}), expression(NH[4]^{"+"}), 
+  expression(NO[3]^{"-"}), "BC", "OC", "Sea salt", "Dust")
 spec_pal <- c(2, 6, 4, 1, 3, 5, 7)
 
 # Zeros imputation
@@ -90,8 +90,10 @@ mean_comp <- t(mean_comp)
 p <- ncol(mean_comp)
 colnames(mean_comp) <- spec_names
 var_mat <- variation(acomp(mean_comp))
+spec_labs[6] <- expression("SS")
 colnames(var_mat) <- rownames(var_mat) <- paste(":", spec_labs) 
 
+
 # Color palette
 colpal <- colorRampPalette(c('#FFFFFF', '#D1E5F0', '#92C5DE',
   '#4393C3', '#2166AC', '#053061'))(20)
@@ -100,5 +102,5 @@ x11()
 corrplot.mixed(var_mat, is.corr = F, tl.col = spec_pal, tl.cex = 1.5,
   lower.col = colpal, upper.col = colpal)
 
-dev.print(pdf, file = "Results/eFigure7.pdf")
+dev.print(png, file = "Results/eFigure7.png", res = 1000, units = "in")
 

3_SecondStage_MetaRegComposition.R

@@ -24,7 +24,7 @@ load("Data/2_FirstStageResults.RData")
 spec_inds <- grep("PM25", colnames(dlist_spec[[1]]))
 spec_names <- c("SO4", "NH4", "NO3", "BC", "OC", "SS", "DUST")
 spec_labs <- c(expression(SO[4]^{"2-"}), expression(NH[4]^{"+"}), 
-  expression(NO[3]^{"-"}), "BC", "OC", "SS", "DUST")
+  expression(NO[3]^{"-"}), "BC", "OC", "Sea salt", "Dust")
 spec_pal <- c(2, 6, 4, 1, 3, 5, 7)
 
 # Mean per city
@@ -208,7 +208,7 @@ pm <- rbind(c(9, rep(10, 3)),
 
 # Initialize plot
 x11(height = 10, width = 15)
-par(mar = c(5, 5, 3, 1), cex.main = 1.5, cex.lab = 1.2)
+par(mar = c(5, 6, 3, 1), cex.main = 1.5, cex.lab = 1.2)
 layout(pm, width = c(.4, .2, .2, .2), height = c(.1, .3, .3, .3))
 
 #----- Panel A: meta-coefficients

4_SupplementaryResults.R

@@ -56,7 +56,7 @@ legend("topright", legend = unique(cities$region),
   inset = .02, ncol = 1, xpd = T
 )
 
-dev.print(pdf, file = "Results/eFigure1.pdf")
+dev.print(png, filename = "Results/eFigure1.png", res = 1000, units = "in")
 
 #-------------------------------------
 #  Supplemental Material B: Residual Analysis
@@ -70,7 +70,7 @@ x11()
 hist(resvec, border = "white", col = 4, xlab = "Residuals", freq = F,
      main = "")
 
-dev.print(pdf, file = "Results/eFigure2.pdf")
+dev.print(png, file = "Results/eFigure2.png", res = 1000, units = "in")
 
 #---- Plot residuals by region ----
 
@@ -97,7 +97,7 @@ abline(h = 0)
 legend(par("usr")[2], par("usr")[4], unique(cities_ord$region), pch = reg_pch,
   col = reg_pal, bty = "n", xpd = T, title = "Region")
 
-dev.print(pdf, file = "Results/eFigure3.pdf")
+dev.print(png, file = "Results/eFigure3.png", res = 1000, units = "in")
 
 #---- Plot residuals by components ----
 pm <- matrix(c(1:6, 0, 7, 0), nrow = 3, byrow = T)
@@ -111,7 +111,7 @@ for (j in seq(p)){
     main = bquote(bold(.(spec_labs[j][[1]]))))
 }
 
-dev.print(pdf, file = "Results/eFigure4.pdf")
+dev.print(png, file = "Results/eFigure4.png", res = 1000, units = "in")
 
 #---- Check extreme residuals ----
 res_sort <- order(resvec)
@@ -141,7 +141,7 @@ legend(par("usr")[2], with(lg$rect, top - h),
   c("Observed range", "Extrapolation"), lwd = c(3, 1), lty = 1:2, bty = "n", 
   xpd = T, y.intersp = 1.2, col = grey(.5))
 
-dev.print(pdf, file = "Results/eFigure5.pdf")
+dev.print(png, file = "Results/eFigure5.png", res = 1000, units = "in")
 
 #----- Plot ternary diagram with predictions -----
 ## Create function to predict for Ternary plot
@@ -210,7 +210,7 @@ image.plot(zlim = c(0.985, 1.015), col = colpal(20),
   legend.only = T, xpd = T, smallplot = c(0, .05, 0.2, .8))
 text(0, .92, "RR", xpd = T, cex = 1.5)
 
-dev.print(pdf, file = "Results/eFigure6.pdf")
+dev.print(png, file = "Results/eFigure6.png", res = 1000, units = "in")
 
 
 #-------------------------------------

README.md

@@ -1,12 +1,17 @@
 # Differential mortality risks associated to PM2.5 components: a multi-country multi-city study.
 
-Reproducible R code and results for the publication 
+R code and results attached to the publication:
 
 Masselot P, et al. Differential mortality risks associated to PM2.5 components: a multi-country multi-city study. *Epidemiology*. In press.
 
 ### Data and Results
 
-Figures and Tables (including supplemental ones) are available in the Results folder of this repository. Data are not available currently due to restricted data sharing agreement between the collaborators of this study.
+**Data are not available currently due to restricted data sharing agreement between the collaborators of this study. Therefore, the code is not fully reproducible.**
+
+Data are normally included in a subfolder *Data*. It should contain:
+- Mortality and pollution data, stored in a list of city-specific data.frames. Also contains a descriptive data.frame with one line for each city.
+- PM2.5 components, stored in one csv files per year of data. Each csv file contains one line per city.
+- City-specific characteristics, stored as a data.frame with one line per city.
 
 ### R code
 
@@ -18,3 +23,7 @@ The R code to reproduce the analysis is available. Scripts are meant to be execu
 - *2_FirstStage.R* Runs the first-stage model on each selected city to produce city-level relative risks.
 - *3_SecondStage_MetaRegComposition.R* Pools first-stage results in a meta analysis including city-specific compositions and socio-economic indicators. Produces Figures 1 and 3 as well as several eFigures.
 - *4_SupplementaryResults.R* Produces several supplementary results and eFigures, including the PCA summary and residuals analysis.
+
+### Results
+
+Figures and Tables generated for the publication (including supplemental ones) are available in the Results folder of this repository.