Mercurial > repos > pravs > protein_rna_correlation
view protein_rna_correlation.r @ 10:c2ac6f10b456 draft default tip
planemo upload
| author | pravs |
|---|---|
| date | Wed, 22 Aug 2018 15:08:45 -0400 |
| parents | e407b1a7a8de |
| children |
line wrap: on
line source
#================================================================================== # About the script #================================================================================== # Version: V1 # This script works for single sample only # It takes GE (Gene Expression) and PE (Protein expression) data of one sample and perform correlation, regression analysis between PE and GE data # Input data can be of tsv format # Script also need a parameter or option file #================================================================================== # Dependencies #================================================================================== # Following R package has to be installed. # data.table # gplots # MASS # DMwR # mgcv # It can be installed by following R command in R session. e.g. install.packages("data.table") #================================================================================== # How to Run #================================================================================== # Rscript PE_GE_association_singleSample_V1.r <PE_file> <GE_file> <Option_file containing tool parameters> <Ensembl map file containing directory path> <outdir> #================================================================================== # Arguments #================================================================================== # Arg1. <PE file>: PE data (tsv format) # Arg2. <GE file>: GE data (tsv format) # Arg3. <Option file>: tsv format, key\tvalue # Options are # PE_idcolno: Column number of PE file containing protein IDs # GE_idcolno: Column number of GE file containing transcript IDs # PE_expcolno: Column number of PE file containing protein expression values # GE_expcolno: Column number of GE file containing transcript expression values # PE_idtype: protein id type. It can be either Uniprot or Ensembl or HGNC_symbol # GE_idtype: transcript id type. At present it is only one type i.e. Ensembl or HGNC_symbol # Organism: Organism # writeMapUnmap: Whether to write mapped and unmapped data in input data format. It takes value as 1 or 0. If 1, mapped and unmapped data is written. Default is 1. # doscale: Whether perform scaling to input data or not. If yet, abundance values are normalized by standard normalization. Default 1 # Arg4. <Ensembl map file containg directory>: Path to Ensembl map file containg directory e.g. /home/user/Ensembl/mapfiles # Arg5. <Outdir>: output directory (e.g. /home/user/out1) #================================================================================== # Sample option file #================================================================================== #PE_idcolno 7 #GE_idcolno 1 #PE_expcolno 2 #GE_expcolno 3 #PE_idtype Ensembl #GE_idtype Ensembl #Organism mmusculus #writeMapUnmap 1 #doscale 1 #================================================================================== # Output #================================================================================== # The script outputs image and data folder along with Correlation_result.html and Result.log file # Result.log: Log file # Correlation_result.html; main result file in html format # data folder contains following output files # PE_abundance.tsv: 2 column tsv file containing mapped id and protein expression values # GE_abundance.tsv: 2 column tsv file containing mapped id and transcript expression values # If writeMapUnmap is 1 i.e. to write mapped and unmapped data, 4 additional file will be written # PE_unmapped.tsv: Output format is same as input, PE unmapped data is written # GE_unmapped.tsv: Output format is same as input, GE unmapped data is written # PE_mapped.tsv: Output format is same as input, PE mapped data is written # GE_mapped.tsv: Output format is same as input, GE mapped data is written # PE_GE_influential_observation.tsv: Influential observations based on cook's distance # PE_GE_kmeans_clusterpoints.txt: Observations clustered based on kmeans clustering. File contains cluster assignment of each observations #================================================================================== # ............................SCRIPT STARTS FROM HERE ............................. #================================================================================== # Warning Off # oldw <- getOption("warn") # options(warn = -1) #============================================================================================================= # Functions #============================================================================================================= usage <- function() { cat("\n\n###########\nERROR: Rscript PE_GE_association_singleSample_V1.r <PE file> <GE file> <Option file containing tool parameters> <Ensembl map file containing directory path> <outdir>\n###########\n\n"); } #============================================================================================================= # Global variables #============================================================================================================= noargs = 13; #============================================================================================================= # Parse command line arguments in args object #============================================================================================================= args = commandArgs(trailingOnly = TRUE); #============================================================================================================= # Check for No of arguments #============================================================================================================= if(length(args) != noargs) { usage(); stop(paste("Please check usage. Number of arguments is not equal to ",noargs,sep="",collapse="")); } #====================================================================== # Load libraries #====================================================================== library(data.table); library(lattice); library(grid); library(nlme); library(gplots); library(MASS); library(DMwR); library(mgcv); #============================================================================================================= # Set variables #============================================================================================================= #PE_file = args[1]; # Protein abundance data file #GE_file = args[2]; # Gene expression data file #option_file = args[3]; # Option file containing various parameters #biomartdir = args[4]; # Biomart map file containing directory path in local system #outdir = args[5]; # output directory PE_file = args[1]; # Protein abundance data file GE_file = args[2]; # Gene expression data file #option_file = args[3]; # Option file containing various parameters #biomartdir = args[4]; # Biomart map file containing directory path in local system outdir = args[13]; # output directory #imagesubdirprefix = "image"; #datasubdirprefix = "data"; #htmloutfile = "Correlation_result.html"; htmloutfile = args[12] #logfile = "Result.log"; PE_outfile_mapped = "PE_mapped.tsv"; GE_outfile_mapped = "GE_mapped.tsv"; PE_outfile_unmapped = "PE_unmapped.tsv"; GE_outfile_unmapped = "GE_unmapped.tsv"; PE_expfile = "PE_abundance.tsv"; GE_expfile = "GE_abundance.tsv"; PE_outfile_excluded_naInf = "PE_excluded_NA_Inf.tsv"; GE_outfile_excluded_naInf = "GE_excluded_NA_Inf.tsv"; #============================================================================================================= # Check input files existance #============================================================================================================= if(! file.exists(PE_file)) { usage(); stop(paste("Input PE_file file does not exists. Path given: ",PE_file,sep="",collapse="")); } if(! file.exists(GE_file)) { usage(); stop(paste("Input GE_file does not exists. Path given: ",GE_file,sep="",collapse="")); } #if(! file.exists(option_file)) #{ # usage(); # stop(paste("Input option_file does not exists. Path given: ",option_file,sep="",collapse="")); #} #============================================================================================================= # Read param_file and set parameter/option data frame #============================================================================================================= #optiondf = read.table(option_file, header = FALSE, stringsAsFactors = FALSE) #rownames(optiondf) = optiondf[,1]; #============================================================================================================= # Define option variables #============================================================================================================= #PE_idcolno = as.numeric(optiondf["PE_idcolno",2]); #GE_idcolno = as.numeric(optiondf["GE_idcolno",2]); #PE_expcolno = as.numeric(optiondf["PE_expcolno",2]); #GE_expcolno = as.numeric(optiondf["GE_expcolno",2]); #PE_idtype = optiondf["PE_idtype",2]; #GE_idtype = optiondf["GE_idtype",2]; #Organism = optiondf["Organism",2]; #writeMapUnmap = as.logical(as.numeric(optiondf["writeMapUnmap",2])); #doscale = as.logical(as.numeric(optiondf["doscale",2])); PE_idcolno = as.numeric(args[3]) GE_idcolno = as.numeric(args[4]) PE_expcolno = as.numeric(args[5]) GE_expcolno = as.numeric(args[6]) PE_idtype = args[7] GE_idtype = args[8] #Organism = args[9] writeMapUnmap = as.logical(as.numeric(args[10])); doscale = as.logical(as.numeric(args[11])); #1 PE_file = "test_data/PE_mouse_singlesample.txt" #2 GE_file = "test_data/GE_mouse_singlesample.txt" #3 PE_idcolno = 7 #4 GE_idcolno = 1 #5 PE_expcolno = 13 #6 GE_expcolno = 10 #7 PE_idtype = "Ensembl_with_version" #8 GE_idtype = "Ensembl_with_version" #10 writeMapUnmap = 1 #11 doscale = 1 #9 biomart_mapfile = "test_data/mmusculus_gene_ensembl__GRCm38.p6.map" #12 htmloutfile = "html_out.html" #13 outdir = "output_elements" #============================================================================================================= # Set column name of biomart map file (idtype) based on whether Ensembl or Uniprot #============================================================================================================= if(PE_idtype == "Ensembl") { PE_idtype = "ensembl_peptide_id"; }else { if(PE_idtype == "Ensembl_with_version") { PE_idtype = "ensembl_peptide_id_version"; }else{ if(PE_idtype == "HGNC_symbol") { PE_idtype = "hgnc_symbol"; } } } if(GE_idtype == "Ensembl") { GE_idtype = "ensembl_transcript_id"; }else { if(GE_idtype == "Ensembl_with_version") { GE_idtype = "ensembl_transcript_id_version"; }else{ if(GE_idtype == "HGNC_symbol") { GE_idtype = "hgnc_symbol"; } } } #============================================================================================================= # Identify biomart mapping file #============================================================================================================= #biomartdir = gsub(biomartdir, pattern="/$", replacement="") #biomart_mapfilename = list.files(path = biomartdir, pattern = Organism); #biomart_mapfile = paste(biomartdir,"/",biomart_mapfilename,sep="",collapse=""); #print(biomart_mapfile); biomart_mapfile = args[9]; #============================================================================================================= # Parse PE, GE, biomart file file #============================================================================================================= PE_df = as.data.frame(fread(input=PE_file, header=T, sep="\t")); GE_df = as.data.frame(fread(input=GE_file, header=T, sep="\t")); biomart_df = as.data.frame(fread(input=biomart_mapfile, header=T, sep="\t")); #============================================================================================================= # Create directory structures and then set the working directory to output directory #============================================================================================================= if(! file.exists(outdir)) { dir.create(outdir); } #tempdir = paste(outdir,"/",imagesubdirprefix,sep="",collapse=""); #if(! file.exists(tempdir)) #{ # dir.create(tempdir); #} #tempdir = paste(outdir,"/",datasubdirprefix,sep="",collapse=""); #if(! file.exists(tempdir)) #{ # dir.create(tempdir); #} #setwd(outdir); logfile = paste(outdir,"/", "Result.log",sep="",collapse=""); #============================================================================================================= # Write initial data summary in html outfile #============================================================================================================= cat("<html><body>\n", file = htmloutfile); cat("<h1>Association between proteomics and transcriptomics data</h1>\n", "<font color='blue'><h3>Input data summary</h3></font>", "<ul>", "<li>Abbrebiations used: PE (Proteomics) and GE (Transcriptomics)","</li>", "<li>Input PE data dimension (Row Column): ", dim(PE_df),"</li>", "<li>Input GE data dimension (Row Column): ", dim(GE_df),"</li>", #"<li>Organism selected: ", Organism,"</li>", "<li>Protein ID fetched from column: ", PE_idcolno,"</li>", "<li>Transcript ID fetched from column: ", GE_idcolno,"</li>", "<li>Protein ID type: ", PE_idtype,"</li>", "<li>Transcript ID type: ", GE_idtype,"</li>", "<li>Protein expression data fetched from column: ", PE_expcolno,"</li>", "<li>Transcript expression data fetched from column: ", GE_expcolno,"</li>", file = htmloutfile, append = TRUE); #============================================================================================================= # Write initial data summary in logfile #============================================================================================================= #cat("Current work dir:", outdir,"\n"); cat("Processing started\n---------------------\n", file=logfile); cat("Abbrebiations used: PE (Proteomics) and GE (Transcriptomics)\n", file=logfile, append=T); cat("Input PE data dimension (Row Column): ", dim(PE_df),"\n", file=logfile, append=T) cat("Input GE data dimension (Row Column): ", dim(GE_df),"\n", file=logfile, append=T) #cat("Organism selected: ", Organism,"\n", file=logfile, append=T) #cat("Biomart map file used: ", biomart_mapfilename,"\n", file=logfile, append=T) cat("Ensembl Biomart mapping data dimension (Row Column): ", dim(biomart_df),"\n", file=logfile, append=T) cat("\n\nProtein ID to Transcript ID mapping started\n----------------\n", file=logfile, append=T) cat("Protein ID fetched from column:", PE_idcolno,"\n", file=logfile, append=T) cat("Transcript ID fetched from column:", GE_idcolno, "\n",file=logfile, append=T) cat("Protein ID type:", PE_idtype, "\n",file=logfile, append=T) cat("Transcript ID type:", GE_idtype,"\n", file=logfile, append=T); cat("Protein expression data fetched from column:", PE_expcolno,"\n", file=logfile, append=T) cat("Transcript expression data fetched from column:", GE_expcolno, "\n",file=logfile, append=T) #============================================================================================================= # Mapping starts here # Pseudocode # Loop over each row of PE file, fetch protein_id # Search the biomartmap file and obtain corresponding transcript_id # Take the mapped transcript_id and search in GE file, which row it correspond # Store the PE rowno and GE rowno in rowpair object #============================================================================================================= rowpair = data.frame(PE_rowno = 0, GE_rowno = 0); cat("Total rows:", nrow(PE_df),"\n"); cat("\n\nTotal protein ids to be mapped: ", nrow(PE_df),"\n", file=logfile, append=T); messagelog = "\n\nBelow are protein IDs, for which no match is observed in Ensembl Biomart Map file.\n\n"; # GE_id column GE_ids = GE_df[,GE_idcolno]; GE_ids = gsub(x=GE_ids, pattern=".\\d+$", replacement=""); # Remove version # Loop over every row of PE data (PE_df) for(i in 1:nrow(PE_df)) { if(i%%100 ==0) { cat("Total rows processed ", i,"\n", file=logfile, append=T); print(i); } queryid = PE_df[i,PE_idcolno]; #queryid = gsub(x=queryid, pattern=".\\d+$", replacement=""); # Remove version #print(queryid); PE_df_matchrowno = i; # Row number in PE_df which matches queryid if(PE_idtype == "Uniprot") { biomart_matchrowno = which(biomart_df[,8] == queryid | biomart_df[,9] == queryid); # Row number of biomart_df which matches queryid }else{ biomart_matchrowno = which(biomart_df[,PE_idtype] == queryid); # Row number of biomart_df which matches queryid } # If match found, map protein id to GE id and find corresponding match row number of GE_df if(length(biomart_matchrowno)>0) { GE_df_matchrowno = which(GE_ids %in% biomart_df[biomart_matchrowno[1],GE_idtype]); rowpair = rbind( rowpair, c(PE_df_matchrowno, GE_df_matchrowno)); if(length(GE_df_matchrowno) > 1) { cat("\nFor protein ID ", i," multiple transcript mapping found\n", file=logfile, append=T); cat( "<br><font color=",'red',">For protein ID", i," multiple transcript mapping found</font><br>",file = htmloutfile, append = TRUE); } }else{ messagelog = paste(messagelog, queryid, "\n",sep="", collapse=""); } } rowpair = rowpair[-1,]; #============================================================================================================= # Write mapping summary, mapped and unmapped data #============================================================================================================= cat( "<li>Total Protein ID mapped: ", length(intersect(1:nrow(PE_df), rowpair[,1])),"</li>", "<li>Total Protein ID unmapped: ", length(setdiff(1:nrow(PE_df), rowpair[,1])),"</li>", "<li>Total Transcript ID mapped: ", length(intersect(1:nrow(GE_df), rowpair[,2])),"</li>", "<li>Total Transcript ID unmapped: ", length(setdiff(1:nrow(GE_df), rowpair[,2])),"</li></ul>", file = htmloutfile, append = TRUE); cat("\n\nMapping Statistics\n---------------------\n", file=logfile, append=T); cat("Total Protein ID mapped:", length(intersect(1:nrow(PE_df), rowpair[,1])), "\n", file=logfile, append=T) cat("Total Protein ID unmapped:", length(setdiff(1:nrow(PE_df), rowpair[,1])), "\n", file=logfile, append=T) cat("Total Transcript ID mapped:", length(intersect(1:nrow(GE_df), rowpair[,2])), "\n", file=logfile, append=T) cat("Total Transcript ID unmapped:", length(setdiff(1:nrow(GE_df), rowpair[,2])), "\n", file=logfile, append=T) cat(messagelog,"\n",file=logfile, append=T); if(writeMapUnmap) { write.table(PE_df[rowpair[,1], ], file=paste(outdir,"/",PE_outfile_mapped,sep="",collapse=""), row.names=F, quote=F, sep="\t", eol="\n") write.table(GE_df[rowpair[,2], ], file= paste(outdir,"/",GE_outfile_mapped,sep="",collapse=""), row.names=F, quote=F, sep="\t", eol="\n") write.table(PE_df[-rowpair[,1], ], file= paste(outdir,"/",PE_outfile_unmapped,sep="",collapse=""), row.names=F, quote=F, sep="\t", eol="\n") write.table(GE_df[-rowpair[,2], ], file=paste(outdir,"/",GE_outfile_unmapped,sep="",collapse=""), row.names=F, quote=F, sep="\t", eol="\n"); cat( "<font color='blue'><h3>Download mapped unmapped data</h3></font>", "<ul><li>Protein mapped data: ", '<a href="',paste(PE_outfile_mapped,sep="",collapse=""),'" target="_blank"> Link</a>',"</li>", "<li>Protein unmapped data: ", '<a href="',paste(PE_outfile_unmapped,sep="",collapse=""),'" target="_blank"> Link</a>',"</li>", "<li>Transcript mapped data: ", '<a href="',paste(GE_outfile_mapped,sep="",collapse=""),'" target="_blank"> Link</a>',"</li>", "<li>Transcript unmapped data: ", '<a href="',paste(GE_outfile_unmapped,sep="",collapse=""),'" target="_blank"> Link</a>',"</li>", file = htmloutfile, append = TRUE); } write.table(PE_df[rowpair[,1], c(PE_idcolno,PE_expcolno)], file=paste(outdir,"/",PE_expfile,sep="",collapse=""), row.names=F, quote=F, sep="\t", eol="\n") write.table(GE_df[rowpair[,2], c(GE_idcolno,GE_expcolno)], file=paste(outdir,"/",GE_expfile,sep="",collapse=""), row.names=F, quote=F, sep="\t", eol="\n") cat( "<li>Protein abundance data: ", '<a href="',paste(PE_expfile,sep="",collapse=""),'" target="_blank"> Link</a>',"</li>", "<li>Transcript abundance data: ", '<a href="',paste(GE_expfile,sep="",collapse=""),'" target="_blank"> Link</a>',"</li></ul>", file = htmloutfile, append = TRUE); #========================================================================================== # Analysis (correlation and regression) starts here #========================================================================================== cat("Analysis started\n---------------------------\n",file=logfile, append=T); proteome_df = PE_df[rowpair[,1], c(PE_idcolno,PE_expcolno)]; transcriptome_df = GE_df[rowpair[,2], c(GE_idcolno,GE_expcolno)]; nPE = nrow(proteome_df); nGE = nrow(transcriptome_df) cat("Total Protein ID: ",nPE,"\n",file=logfile, append=T); cat("Total Transcript ID: ",nGE,"\n",file=logfile, append=T); #========================================================================================== # Data summary #========================================================================================== cat( "<ul>", "<li>Number of entries in Transcriptome data used for correlation: ",nPE,"</li>", "<li>Number of entries in Proteome data used for correlation: ",nGE,"</li></ul>", file = htmloutfile, append = TRUE); #============================================================================================================= # Remove entries with NA or Inf or -Inf in Transcriptome and Proteome data which will create problem in correlation analysis #============================================================================================================= totna = sum(is.na(transcriptome_df[,2]) | is.na(proteome_df[,2])); totinf = sum(is.infinite(transcriptome_df[,2]) | is.infinite(proteome_df[,2])); cat("<font color='blue'><h3>Filtering</h3></font>","Checking for NA or Inf or -Inf in either Transcriptome or Proteome data, if found, remove those entry<br>","<ul>","<li>Number of NA found: ",totna,"</li>","<li>Number of Inf or -Inf found: ",totinf,"</li></ul>",file = htmloutfile, append = TRUE); cat("Total NA observed in either Transcriptome or Proteome data: ",totna,"\n",file=logfile, append=T); cat("Total Inf or -Inf observed in either Transcriptome or Proteome data: ",totinf,"\n",file=logfile, append=T); if(totna > 0 | totinf > 0) { excludeIndices_PE = which(is.na(proteome_df[,2]) | is.infinite(proteome_df[,2])); excludeIndices_GE = which(is.na(transcriptome_df[,2]) | is.infinite(transcriptome_df[,2])); excludeIndices = which(is.na(transcriptome_df[,2]) | is.infinite(transcriptome_df[,2]) | is.na(proteome_df[,2]) | is.infinite(proteome_df[,2])); # Write excluded transcriptomics and proteomics data to file write.table(proteome_df[excludeIndices_PE,], file=paste(outdir,"/",PE_outfile_excluded_naInf,sep="",collapse=""), row.names=F, quote=F, sep="\t", eol="\n") write.table(transcriptome_df[excludeIndices_GE,], file=paste(outdir,"/",GE_outfile_excluded_naInf,sep="",collapse=""), row.names=F, quote=F, sep="\t", eol="\n") # Write excluded transcriptomics and proteomics data link to html file cat( "<ul><li>Protein excluded data with NA or Inf or -Inf: ", '<a href="',paste(PE_outfile_excluded_naInf,sep="",collapse=""),'" target="_blank"> Link</a>',"</li>", "<li>Transcript excluded data with NA or Inf or -Inf: ", '<a href="',paste(GE_outfile_excluded_naInf,sep="",collapse=""),'" target="_blank"> Link</a>',"</li></ul>", file = htmloutfile, append = TRUE); # Keep the unexcluded entries only transcriptome_df = transcriptome_df[-excludeIndices,]; proteome_df = proteome_df[-excludeIndices,]; nPE = nrow(proteome_df); nGE = nrow(transcriptome_df) cat("<font color='blue'><h3>Filtered data summary</h3></font>", "Excluding entires with abundance values: NA/Inf/-Inf<br>", "<ul>", "<li>Number of entries in Transcriptome data remained: ",nrow(transcriptome_df),"</li>", "<li>Number of entries in Proteome data remained: ",nrow(proteome_df),"</li></ul>", file = htmloutfile, append = TRUE); cat("Excluding entires with abundance values: NA/Inf/-Inf","\n",file=logfile, append=T); cat("Total Protein ID after filtering: ",nPE,"\n",file=logfile, append=T); cat("Total Transcript ID after filtering: ",nGE,"\n",file=logfile, append=T); } #========================================================================================== # Scaling of data #========================================================================================== if(doscale) { proteome_df[,2] = scale(proteome_df[,2], center = TRUE, scale = TRUE); transcriptome_df[,2] = scale(transcriptome_df[,2], center = TRUE, scale = TRUE); } #============================================================================================================= # Proteome and Transcriptome data summary #============================================================================================================= cat("Calculating summary of PE and GE data\n",file=logfile, append=T); s1 = summary(proteome_df[,2]); s2 = summary(transcriptome_df[,2]) cat( "<font color='blue'><h3>Proteome data summary</h3></font>\n", '<table class="embedded-table" border=1 cellspacing=0 cellpadding=5 style="table-layout:auto; "> <tr bgcolor="#c3f0d6"><th>Parameter</th><th>Value</th></tr>', file = htmloutfile, append = TRUE); for(i in 1:length(s1)) { cat("<tr><td>",names(s1[i]),"</td><td>", s1[i],"</td></tr>\n", file = htmloutfile, append = TRUE) } cat("</table>\n", file = htmloutfile, append = TRUE) cat( "<font color='blue'><h3>Transcriptome data summary</h3></font>\n", '<table class="embedded-table" border=1 cellspacing=0 cellpadding=5 style="table-layout:auto; "> <tr bgcolor="#c3f0d6"><th>Parameter</th><th>Value</th></tr>', file = htmloutfile, append = TRUE); for(i in 1:length(s2)) { cat("<tr><td>",names(s2[i]),"</td><td>", s2[i],"</td></tr>\n", file = htmloutfile, append = TRUE) } cat("</table>\n", file = htmloutfile, append = TRUE) #============================================================================================================= # Distribution of proteome and transcriptome abundance (Box and Density plot) #============================================================================================================= cat("Generating Box and Density plot\n",file=logfile, append=T); outplot = paste(outdir,"/AbundancePlot.png",sep="",collapse=""); #png(outplot); bitmap(outplot, "png16m"); par(mfrow=c(2,2)); boxplot(proteome_df[,2], ylab="Abundance", main="Proteome abundance", cex.lab=1.5); plot(density(proteome_df[,2]), xlab="Protein Abundance", ylab="Density", main="Proteome abundance", cex.lab=1.5); boxplot(transcriptome_df[,2], ylab="Abundance", main="Transcriptome abundance", cex.lab=1.5); plot(density(transcriptome_df[,2]), xlab="Transcript Abundance", ylab="Density", main="Transcriptome abundance", cex.lab=1.5); dev.off(); cat( "<font color='blue'><h3>Distribution of Proteome and Transcripome abundance (Box plot and Density plot)</h3></font>\n", '<img src="AbundancePlot.png">', file = htmloutfile, append = TRUE); #============================================================================================================= # Scatter plot #============================================================================================================= cat("Generating scatter plot\n",file=logfile, append=T); outplot = paste(outdir,"/AbundancePlot_scatter.png",sep="",collapse=""); #png(outplot); bitmap(outplot,"png16m") par(mfrow=c(1,1)); scatter.smooth(transcriptome_df[,2], proteome_df[,2], xlab="Transcript Abundance", ylab="Protein Abundance", cex.lab=1.5); cat( "<font color='blue'><h3>Scatter plot between Proteome and Transcriptome Abundance</h3></font>\n", '<img src="AbundancePlot_scatter.png">', file = htmloutfile, append = TRUE); #============================================================================================================= # Correlation testing #============================================================================================================= cat("Estimating correlation\n",file=logfile, append=T); cor_result_pearson = cor.test(transcriptome_df[,2], proteome_df[,2], method = "pearson"); cor_result_spearman = cor.test(transcriptome_df[,2], proteome_df[,2], method = "spearman"); cor_result_kendall = cor.test(transcriptome_df[,2], proteome_df[,2], method = "kendall"); cat( "<font color='blue'><h3>Correlation with all data</h3></font>\n", '<table class="embedded-table" border=1 cellspacing=0 cellpadding=5 style="table-layout:auto; "> <tr bgcolor="#c3f0d6"><th>Parameter</th><th>Method 1</th><th>Method 2</th><th>Method 3</th></tr>', file = htmloutfile, append = TRUE); cat( "<tr><td>Correlation method used</td><td>",cor_result_pearson$method,"</td><td>",cor_result_spearman$method,"</td><td>",cor_result_kendall$method,"</td></tr>", "<tr><td>Correlation</td><td>",cor_result_pearson$estimate,"</td><td>",cor_result_spearman$estimate,"</td><td>",cor_result_kendall$estimate,"</td></tr>", "<tr><td>Pvalue</td><td>",cor_result_pearson$p.value,"</td><td>",cor_result_spearman$p.value,"</td><td>",cor_result_kendall$p.value,"</td></tr>", file = htmloutfile, append = TRUE) cat("</table>\n", file = htmloutfile, append = TRUE) cat( '<font color="red">*Note that <u>correlation</u> is <u>sensitive to outliers</u> in the data. So it is important to analyze outliers/influential observations in the data.<br> Below we use <u>cook\'s distance based approach</u> to identify such influential observations.</font>', file = htmloutfile, append = TRUE); #============================================================================================================= # Linear Regression #============================================================================================================= cat("Fitting linear regression model\n",file=logfile, append=T); PE_GE_data = proteome_df; PE_GE_data = cbind(PE_GE_data, transcriptome_df); colnames(PE_GE_data) = c("PE_ID","PE_abundance","GE_ID","GE_abundance"); regmodel = lm(PE_abundance~GE_abundance, data=PE_GE_data); regmodel_summary = summary(regmodel); cat( "<font color='blue'><h3>Linear Regression model fit between Proteome and Transcriptome data</h3></font>\n", "<p>Assuming a linear relationship between Proteome and Transcriptome data, we here fit a linear regression model.</p>\n", '<table class="embedded-table" border=1 cellspacing=0 cellpadding=5 style="table-layout:auto; "> <tr bgcolor="#c3f0d6"><th>Parameter</th><th>Value</th></tr>', file = htmloutfile, append = TRUE); cat( "<tr><td>Formula</td><td>","PE_abundance~GE_abundance","</td></tr>\n", "<tr><td colspan='2' align='center'> <b>Coefficients</b></td>","</tr>\n", "<tr><td>",names(regmodel$coefficients[1]),"</td><td>",regmodel$coefficients[1]," (Pvalue:", regmodel_summary$coefficients[1,4],")","</td></tr>\n", "<tr><td>",names(regmodel$coefficients[2]),"</td><td>",regmodel$coefficients[2]," (Pvalue:", regmodel_summary$coefficients[2,4],")","</td></tr>\n", "<tr><td colspan='2' align='center'> <b>Model parameters</b></td>","</tr>\n", "<tr><td>Residual standard error</td><td>",regmodel_summary$sigma," (",regmodel_summary$df[2]," degree of freedom)</td></tr>\n", "<tr><td>F-statistic</td><td>",regmodel_summary$fstatistic[1]," ( on ",regmodel_summary$fstatistic[2]," and ",regmodel_summary$fstatistic[3]," degree of freedom)</td></tr>\n", "<tr><td>R-squared</td><td>",regmodel_summary$r.squared,"</td></tr>\n", "<tr><td>Adjusted R-squared</td><td>",regmodel_summary$adj.r.squared,"</td></tr>\n", file = htmloutfile, append = TRUE) cat("</table>\n", file = htmloutfile, append = TRUE) #============================================================================================================= # Plotting various regression diagnostics plots #============================================================================================================= outplot1 = paste(outdir,"/PE_GE_modelfit.pdf",sep="",collapse=""); pdf(outplot1); devnum = which(unlist(sapply(2:length(.Devices), function(x){attributes(.Devices[[x]])$filepath==outplot1})))+1 print(.Devices) print(c(devnum,"+++")); regmodel_predictedy = predict(regmodel, PE_GE_data); plot(PE_GE_data[,"GE_abundance"], PE_GE_data[,"PE_abundance"], xlab="GE_abundance", ylab="PE_abundance",main="Linear regression with all data"); points(PE_GE_data[,"GE_abundance"], regmodel_predictedy, col="red"); cat("Generating regression diagnostics plot\n",file=logfile, append=T); cat( "<font color='blue'><h3>Plotting various regression diagnostics plots</h3></font>\n", file = htmloutfile, append = TRUE); outplot = paste(outdir,"/PE_GE_lm_1.png",sep="",collapse=""); #png(outplot); bitmap(outplot,"png16m"); par(mfrow=c(1,1)); plot(regmodel, 1, cex.lab=1.5); dev.off(); outplot = paste(outdir,"/PE_GE_lm_2.png",sep="",collapse=""); #png(outplot); bitmap(outplot,"png16m"); par(mfrow=c(1,1)); plot(regmodel, 2, cex.lab=1.5); dev.off(); outplot = paste(outdir,"/PE_GE_lm_3.png",sep="",collapse=""); #png(outplot); bitmap(outplot,"png16m"); par(mfrow=c(1,1)); plot(regmodel, 3, cex.lab=1.5); dev.off(); outplot = paste(outdir,"/PE_GE_lm_5.png",sep="",collapse=""); #png(outplot); bitmap(outplot,"png16m"); par(mfrow=c(1,1)); plot(regmodel, 5, cex.lab=1.5); dev.off(); outplot = paste(outdir,"/PE_GE_lm.pdf",sep="",collapse=""); pdf(outplot); plot(regmodel); dev.off(); regmodel_metrics = regr.eval(PE_GE_data$PE_abundance, regmodel$fitted.values) cat( "<u><font color='brown'><h4>Residuals vs Fitted plot</h4></font></u>\n", '<img src="PE_GE_lm_1.png">', '<br><br>This plot checks for linear relationship assumptions. If a horizontal line is observed without any distinct patterns, it indicates a linear relationship<br>', file = htmloutfile, append = TRUE); cat( "<u><font color='brown'><h4>Normal Q-Q plot of residuals</h4></font></u>\n", '<img src="PE_GE_lm_2.png">', '<br><br>This plot checks whether residuals are normally distributed or not. It is good if the residuals points follow the straight dashed line i.e., do not deviate much from dashed line.<br>', file = htmloutfile, append = TRUE); cat( "<u><font color='brown'><h4>Scale-Location (or Spread-Location) plot</h4></font></u>\n", '<img src="PE_GE_lm_3.png">', '<br><br>This plot checks for homogeneity of residual variance (homoscedasticity). A horizontal line observed with equally spread residual points is a good indication of homoscedasticity.<br>', file = htmloutfile, append = TRUE); cat( "<u><font color='brown'><h4>Residuals vs Leverage plot</h4></font></u>\n", '<img src="PE_GE_lm_3.png">', '<br><br>This plot is useful to identify any influential cases, that is outliers or extreme values that might influence the regression results upon inclusion or exclusion from the analysis.<br>', file = htmloutfile, append = TRUE); #============================================================================================================= # Identification of influential observations #============================================================================================================= cat("Identifying influential observations\n",file=logfile, append=T); cat( "<font color='blue'><h3>Identify influential observations</h3></font>\n", file = htmloutfile, append = TRUE); cat( '<p><b>Cook’s distance</b> computes the influence of each data point/observation on the predicted outcome. i.e. this measures how much the observation is influencing the fitted values.<br>In general use, those observations that have a <b>cook’s distance > than 4 times the mean</b> may be classified as <b>influential.</b></p>', file = htmloutfile, append = TRUE); cooksd <- cooks.distance(regmodel); cat("Generating cooksd plot\n",file=logfile, append=T); outplot = paste(outdir,"/PE_GE_lm_cooksd.png",sep="",collapse=""); #png(outplot); bitmap(outplot,"png16m"); par(mfrow=c(1,1)); plot(cooksd, pch="*", cex=2, cex.lab=1.5,main="Influential Obs. by Cooks distance", ylab="Cook\'s distance", xlab="Observations") # plot cooks distance abline(h = 4*mean(cooksd, na.rm=T), col="red") # add cutoff line #text(x=1:length(cooksd)+1, y=cooksd, labels=ifelse(cooksd>4*mean(cooksd, na.rm=T),names(cooksd),""), col="red", pos=2) # add labels dev.off(); cat( '<img src="PE_GE_lm_cooksd.png">', '<br>In the above plot, observations above red line (4*mean cook\'s distance) are influential, marked in <b>*</b>. Genes that are outliers could be important. These observations influences the correlation values and regression coefficients<br><br>', file = htmloutfile, append = TRUE); tempind = which(cooksd>4*mean(cooksd, na.rm=T)); PE_GE_data_no_outlier = PE_GE_data[-tempind,]; PE_GE_data_no_outlier$cooksd = cooksd[-tempind] PE_GE_data_outlier = PE_GE_data[tempind,]; PE_GE_data_outlier$cooksd = cooksd[tempind] cat( '<table class="embedded-table" border=1 cellspacing=0 cellpadding=5 style="table-layout:auto; "> <tr bgcolor="#c3f0d6"><th>Parameter</th><th>Value</th></tr>', file = htmloutfile, append = TRUE); cat( "<tr><td>Mean cook\'s distance</td><td>",mean(cooksd, na.rm=T),"</td></tr>\n", "<tr><td>Total influential observations (cook\'s distance > 4 * mean cook\'s distance)</td><td>",length(tempind),"</td></tr>\n", "<tr><td>Total influential observations (cook\'s distance > 3 * mean cook\'s distance)</td><td>",length(which(cooksd>3*mean(cooksd, na.rm=T))),"</td></tr>\n", "</table>", '<font color="brown"><h4>Top 10 influential observations (cook\'s distance > 4 * mean cook\'s distance)</h4></font>', file = htmloutfile, append = TRUE); cat("Writing influential observations\n",file=logfile, append=T); outdatafile = paste(outdir,"/PE_GE_influential_observation.tsv", sep="", collapse=""); cat('<a href="',outdatafile, '" target="_blank">Download entire list</a>',file = htmloutfile, append = TRUE); write.table(PE_GE_data_outlier, file=outdatafile, row.names=F, sep="\t", quote=F); cat( '<table class="embedded-table" border=1 cellspacing=0 cellpadding=5 style="table-layout:auto; "> <tr bgcolor="#c3f0d6"><th>PE_ID</th><th>PE_abundance</th><th>GE_ID</th><th>GE_abundance</th><th>cooksd</th></tr>', file = htmloutfile, append = TRUE); for(i in 1:10) { cat( '<tr>','<td>',PE_GE_data_outlier[i,1],'</td>', '<td>',PE_GE_data_outlier[i,2],'</td>', '<td>',PE_GE_data_outlier[i,3],'</td>', '<td>',PE_GE_data_outlier[i,4],'</td>', '<td>',PE_GE_data_outlier[i,5],'</td></tr>', file = htmloutfile, append = TRUE); } cat('</table>',file = htmloutfile, append = TRUE); #============================================================================================================= # Correlation with removal of outliers #============================================================================================================= #============================================================================================================= # Scatter plot #============================================================================================================= outplot = paste(outdir,"/AbundancePlot_scatter_without_outliers.png",sep="",collapse=""); #png(outplot); bitmap(outplot,"png16m"); par(mfrow=c(1,1)); scatter.smooth(PE_GE_data_no_outlier[,"GE_abundance"], PE_GE_data_no_outlier[,"PE_abundance"], xlab="Transcript Abundance", ylab="Protein Abundance", cex.lab=1.5); cat( "<font color='blue'><h3>Scatter plot between Proteome and Transcriptome Abundance, after removal of outliers/influential observations</h3></font>\n", '<img src="AbundancePlot_scatter_without_outliers.png">', file = htmloutfile, append = TRUE); #============================================================================================================= # Correlation #============================================================================================================= cat("Estimating orrelation with removal of outliers \n",file=logfile, append=T); cat( "<font color='blue'><h3>Correlation with removal of outliers / influential observations</h3></font>\n", '<p>We removed the influential observations and reestimated the correlation values.</p>', file = htmloutfile, append = TRUE); cor_result_pearson = cor.test(PE_GE_data_no_outlier[,"GE_abundance"], PE_GE_data_no_outlier[,"PE_abundance"], method = "pearson"); cor_result_spearman = cor.test(PE_GE_data_no_outlier[,"GE_abundance"], PE_GE_data_no_outlier[,"PE_abundance"], method = "spearman"); cor_result_kendall = cor.test(PE_GE_data_no_outlier[,"GE_abundance"], PE_GE_data_no_outlier[,"PE_abundance"], method = "kendall"); cat( '<table class="embedded-table" border=1 cellspacing=0 cellpadding=5 style="table-layout:auto; "> <tr bgcolor="#c3f0d6"><th>Parameter</th><th>Method 1</th><th>Method 2</th><th>Method 3</th></tr>', file = htmloutfile, append = TRUE); cat( "<tr><td>Correlation method used</td><td>",cor_result_pearson$method,"</td><td>",cor_result_spearman$method,"</td><td>",cor_result_kendall$method,"</td></tr>", "<tr><td>Correlation</td><td>",cor_result_pearson$estimate,"</td><td>",cor_result_spearman$estimate,"</td><td>",cor_result_kendall$estimate,"</td></tr>", "<tr><td>Pvalue</td><td>",cor_result_pearson$p.value,"</td><td>",cor_result_spearman$p.value,"</td><td>",cor_result_kendall$p.value,"</td></tr>", file = htmloutfile, append = TRUE) cat("</table>\n", file = htmloutfile, append = TRUE) #============================================================================================================= # Heatmap #============================================================================================================= cat( "<font color='blue'><h3>Heatmap of PE and GE abundance values</h3></font>\n", file = htmloutfile, append = TRUE); cat("Generating heatmap plot\n",file=logfile, append=T); outplot = paste(outdir,"/PE_GE_heatmap.png",sep="",collapse=""); #png(outplot); bitmap(outplot,"png16m"); par(mfrow=c(1,1)); #heatmap.2(as.matrix(PE_GE_data[,c("PE_abundance","GE_abundance")]), trace="none", cexCol=1, col=greenred(100),Colv=F, labCol=c("PE","GE"), scale="col"); my_palette <- colorRampPalette(c("green", "white", "red"))(299); heatmap.2(as.matrix(PE_GE_data[,c("PE_abundance","GE_abundance")]), trace="none", cexCol=1, col=my_palette ,Colv=F, labCol=c("PE","GE"), scale="col", dendrogram = "row"); dev.off(); cat( '<img src="PE_GE_heatmap.png">', file = htmloutfile, append = TRUE); #============================================================================================================= # kmeans clustering #============================================================================================================= PE_GE_data_kdata = PE_GE_data; k1 = kmeans(PE_GE_data_kdata[,c("PE_abundance","GE_abundance")], 5); cat("Generating kmeans plot\n",file=logfile, append=T); outplot = paste(outdir,"/PE_GE_kmeans.png",sep="",collapse=""); #png(outplot); bitmap(outplot,"png16m"); par(mfrow=c(1,1)); scatter.smooth(PE_GE_data_kdata[,"GE_abundance"], PE_GE_data_kdata[,"PE_abundance"], xlab="Transcript Abundance", ylab="Protein Abundance", cex.lab=1.5); ind=which(k1$cluster==1); points(PE_GE_data_kdata[ind,"GE_abundance"], PE_GE_data_kdata[ind,"PE_abundance"], col="red", pch=16); ind=which(k1$cluster==2); points(PE_GE_data_kdata[ind,"GE_abundance"], PE_GE_data_kdata[ind,"PE_abundance"], col="green", pch=16); ind=which(k1$cluster==3); points(PE_GE_data_kdata[ind,"GE_abundance"], PE_GE_data_kdata[ind,"PE_abundance"], col="blue", pch=16); ind=which(k1$cluster==4); points(PE_GE_data_kdata[ind,"GE_abundance"], PE_GE_data_kdata[ind,"PE_abundance"], col="cyan", pch=16); ind=which(k1$cluster==5); points(PE_GE_data_kdata[ind,"GE_abundance"], PE_GE_data_kdata[ind,"PE_abundance"], col="black", pch=16); dev.off(); cat( "<font color='blue'><h3>Kmean clustering</h3></font>\nNumber of Clusters: 5<br>", file = htmloutfile, append = TRUE); tempind = order(k1$cluster); tempoutfile = paste(outdir,"/","PE_GE_kmeans_clusterpoints.txt",sep="",collapse=""); write.table(data.frame(PE_GE_data_kdata[tempind, ], Cluster=k1$cluster[tempind]), file=tempoutfile, row.names=F, quote=F, sep="\t", eol="\n") cat('<a href="',tempoutfile, '" target="_blank">Download cluster list</a><br>',file = htmloutfile, append = TRUE); cat( '<img src="PE_GE_kmeans.png">', file = htmloutfile, append = TRUE); #============================================================================================================= # Other Regression fit #============================================================================================================= dev.set(devnum); # Linear regression with removal of outliers regmodel_no_outlier = lm(PE_abundance~GE_abundance, data=PE_GE_data_no_outlier); regmodel_no_outlier_predictedy = predict(regmodel_no_outlier, PE_GE_data_no_outlier); outplot = paste(outdir,"/PE_GE_lm_without_outliers.pdf",sep="",collapse=""); plot(PE_GE_data_no_outlier[,"GE_abundance"], PE_GE_data_no_outlier[,"PE_abundance"], xlab="GE_abundance", ylab="PE_abundance",main="Linear regression with removal of outliers"); points(PE_GE_data_no_outlier[,"GE_abundance"], regmodel_no_outlier_predictedy, col="red"); pdf(outplot); plot(regmodel_no_outlier); dev.off(); regmodel_no_outlier_metrics = regr.eval(PE_GE_data$PE_abundance, regmodel_no_outlier$fitted.values) # Resistant regression (lqs / least trimmed squares method) regmodel_lqs = lqs(PE_abundance~GE_abundance, data=PE_GE_data); regmodel_lqs_predictedy = predict(regmodel_lqs, PE_GE_data); outplot = paste(outdir,"/PE_GE_lqs.pdf",sep="",collapse=""); pdf(outplot); plot(PE_GE_data[,"GE_abundance"], PE_GE_data[,"PE_abundance"], xlab="GE_abundance", ylab="PE_abundance",main="Resistant regression (lqs / least trimmed squares method)"); points(PE_GE_data[,"GE_abundance"], regmodel_lqs_predictedy, col="red"); #plot(regmodel_lqs); dev.off(); dev.set(devnum); plot(PE_GE_data[,"GE_abundance"], PE_GE_data[,"PE_abundance"], xlab="GE_abundance", ylab="PE_abundance",main="Resistant regression (lqs / least trimmed squares method)"); points(PE_GE_data[,"GE_abundance"], regmodel_lqs_predictedy, col="red"); regmodel_lqs_metrics = regr.eval(PE_GE_data$PE_abundance, regmodel_lqs$fitted.values) # Robust regression (rlm / Huber M-estimator method) regmodel_rlm = rlm(PE_abundance~GE_abundance, data=PE_GE_data); regmodel_rlm_predictedy = predict(regmodel_rlm, PE_GE_data); outplot = paste(outdir,"/PE_GE_rlm.pdf",sep="",collapse=""); plot(PE_GE_data[,"GE_abundance"], PE_GE_data[,"PE_abundance"], xlab="GE_abundance", ylab="PE_abundance",main="Robust regression (rlm / Huber M-estimator method)"); points(PE_GE_data[,"GE_abundance"], regmodel_rlm_predictedy, col="red"); pdf(outplot); plot(regmodel_rlm); dev.off(); regmodel_rlm_metrics = regr.eval(PE_GE_data$PE_abundance, regmodel_rlm$fitted.values) # polynomical reg regmodel_poly2 = lm(PE_abundance ~ poly(GE_abundance, 2, raw = TRUE), data = PE_GE_data) regmodel_poly3 = lm(PE_abundance ~ poly(GE_abundance, 3, raw = TRUE), data = PE_GE_data) regmodel_poly4 = lm(PE_abundance ~ poly(GE_abundance, 4, raw = TRUE), data = PE_GE_data) regmodel_poly5 = lm(PE_abundance ~ poly(GE_abundance, 5, raw = TRUE), data = PE_GE_data) regmodel_poly6 = lm(PE_abundance ~ poly(GE_abundance, 6, raw = TRUE), data = PE_GE_data) regmodel_poly2_metrics = regr.eval(PE_GE_data$PE_abundance, regmodel_poly2$fitted.values) regmodel_poly3_metrics = regr.eval(PE_GE_data$PE_abundance, regmodel_poly3$fitted.values) regmodel_poly4_metrics = regr.eval(PE_GE_data$PE_abundance, regmodel_poly4$fitted.values) regmodel_poly5_metrics = regr.eval(PE_GE_data$PE_abundance, regmodel_poly5$fitted.values) regmodel_poly6_metrics = regr.eval(PE_GE_data$PE_abundance, regmodel_poly6$fitted.values) regmodel_poly2_predictedy = predict(regmodel_poly2, PE_GE_data); regmodel_poly3_predictedy = predict(regmodel_poly3, PE_GE_data); regmodel_poly4_predictedy = predict(regmodel_poly4, PE_GE_data); regmodel_poly5_predictedy = predict(regmodel_poly5, PE_GE_data); regmodel_poly6_predictedy = predict(regmodel_poly6, PE_GE_data); outplot = paste(outdir,"/PE_GE_poly2.pdf",sep="",collapse=""); dev.set(devnum); plot(PE_GE_data[,"GE_abundance"], PE_GE_data[,"PE_abundance"], xlab="GE_abundance", ylab="PE_abundance",main="Polynomial regression with degree 2"); points(PE_GE_data[,"GE_abundance"], regmodel_poly2_predictedy, col="red"); pdf(outplot); plot(regmodel_poly2); dev.off(); outplot = paste(outdir,"/PE_GE_poly3.pdf",sep="",collapse=""); dev.set(devnum); plot(PE_GE_data[,"GE_abundance"], PE_GE_data[,"PE_abundance"], xlab="GE_abundance", ylab="PE_abundance",main="Polynomial regression with degree 3"); points(PE_GE_data[,"GE_abundance"], regmodel_poly3_predictedy, col="red"); pdf(outplot); plot(regmodel_poly3); dev.off(); outplot = paste(outdir,"/PE_GE_poly4.pdf",sep="",collapse=""); dev.set(devnum); plot(PE_GE_data[,"GE_abundance"], PE_GE_data[,"PE_abundance"], xlab="GE_abundance", ylab="PE_abundance",main="Polynomial regression with degree 4"); points(PE_GE_data[,"GE_abundance"], regmodel_poly4_predictedy, col="red"); pdf(outplot); plot(regmodel_poly4); dev.off(); outplot = paste(outdir,"/PE_GE_poly5.pdf",sep="",collapse=""); dev.set(devnum); plot(PE_GE_data[,"GE_abundance"], PE_GE_data[,"PE_abundance"], xlab="GE_abundance", ylab="PE_abundance",main="Polynomial regression with degree 5"); points(PE_GE_data[,"GE_abundance"], regmodel_poly5_predictedy, col="red"); pdf(outplot); plot(regmodel_poly5); dev.off(); outplot = paste(outdir,"/PE_GE_poly6.pdf",sep="",collapse=""); dev.set(devnum); plot(PE_GE_data[,"GE_abundance"], PE_GE_data[,"PE_abundance"], xlab="GE_abundance", ylab="PE_abundance",main="Polynomial regression with degree 6"); points(PE_GE_data[,"GE_abundance"], regmodel_poly6_predictedy, col="red"); pdf(outplot); plot(regmodel_poly6); dev.off(); # GAM Generalized additive models regmodel_gam <- gam(PE_abundance ~ s(GE_abundance), data = PE_GE_data) regmodel_gam_predictedy = predict(regmodel_gam, PE_GE_data); regmodel_gam_metrics = regr.eval(PE_GE_data$PE_abundance, regmodel_gam$fitted.values) outplot = paste(outdir,"/PE_GE_gam.pdf",sep="",collapse=""); dev.set(devnum); plot(PE_GE_data[,"GE_abundance"], PE_GE_data[,"PE_abundance"], xlab="GE_abundance", ylab="PE_abundance",main="Generalized additive models"); points(PE_GE_data[,"GE_abundance"], regmodel_gam_predictedy, col="red"); pdf(outplot); plot(regmodel_gam,pages=1,residuals=TRUE); ## show partial residuals plot(regmodel_gam,pages=1,seWithMean=TRUE) ## `with intercept' CIs dev.off(); dev.off(devnum); cat( "<font color='blue'><h3>Other regression model fitting</h3></font>\n", file = htmloutfile, append = TRUE); cat( "<ul> <li>MAE:mean absolute error</li> <li>MSE: mean squared error</li> <li>RMSE:root mean squared error ( sqrt(MSE) )</li> <li>MAPE:mean absolute percentage error</li> </ul> ", file = htmloutfile, append = TRUE); cat( '<h4><a href="PE_GE_modelfit.pdf" target="_blank">Comparison of model fits</a></h4>', file = htmloutfile, append = TRUE); cat( '<table class="embedded-table" border=1 cellspacing=0 cellpadding=5 style="table-layout:auto; "> <tr bgcolor="#c3f0d6"><th>Model</th><th>MAE</th><th>MSE</th><th>RMSE</th><th>MAPE</th><th>Diagnostics Plot</th></tr>', file = htmloutfile, append = TRUE); cat( "<tr><td>Linear regression with all data</td><td>",regmodel_metrics[1],"</td><td>",regmodel_metrics[2],"</td><td>",regmodel_metrics[3],"</td><td>",regmodel_metrics[4],"</td><td>",'<a href="PE_GE_lm.pdf" target="_blank">Link</a>',"</td></tr>", "<tr><td>Linear regression with removal of outliers</td><td>",regmodel_no_outlier_metrics[1],"</td><td>",regmodel_no_outlier_metrics[2],"</td><td>",regmodel_no_outlier_metrics[3],"</td><td>",regmodel_no_outlier_metrics[4],"</td><td>",'<a href="PE_GE_lm_without_outliers.pdf" target="_blank">Link</a>',"</td></tr>", "<tr><td>Resistant regression (lqs / least trimmed squares method)</td><td>",regmodel_lqs_metrics[1],"</td><td>",regmodel_lqs_metrics[2],"</td><td>",regmodel_lqs_metrics[3],"</td><td>",regmodel_lqs_metrics[4],"</td><td>", '<a href="PE_GE_lqs.pdf" target="_blank">Link</a>',"</td></tr>", "<tr><td>Robust regression (rlm / Huber M-estimator method)</td><td>",regmodel_rlm_metrics[1],"</td><td>",regmodel_rlm_metrics[2],"</td><td>",regmodel_rlm_metrics[3],"</td><td>",regmodel_rlm_metrics[4],"</td><td>",'<a href="PE_GE_rlm.pdf" target="_blank">Link</a>',"</td></tr>", "<tr><td>Polynomial regression with degree 2</td><td>",regmodel_poly2_metrics[1],"</td><td>",regmodel_poly2_metrics[2],"</td><td>",regmodel_poly2_metrics[3],"</td><td>",regmodel_poly2_metrics[4],"</td><td>",'<a href="PE_GE_poly2.pdf" target="_blank">Link</a>',"</td></tr>", "<tr><td>Polynomial regression with degree 3</td><td>",regmodel_poly3_metrics[1],"</td><td>",regmodel_poly3_metrics[2],"</td><td>",regmodel_poly3_metrics[3],"</td><td>",regmodel_poly3_metrics[4],"</td><td>",'<a href="PE_GE_poly3.pdf" target="_blank">Link</a>',"</td></tr>", "<tr><td>Polynomial regression with degree 4</td><td>",regmodel_poly4_metrics[1],"</td><td>",regmodel_poly4_metrics[2],"</td><td>",regmodel_poly4_metrics[3],"</td><td>",regmodel_poly4_metrics[4],"</td><td>",'<a href="PE_GE_poly4.pdf" target="_blank">Link</a>',"</td></tr>", "<tr><td>Polynomial regression with degree 5</td><td>",regmodel_poly5_metrics[1],"</td><td>",regmodel_poly5_metrics[2],"</td><td>",regmodel_poly5_metrics[3],"</td><td>",regmodel_poly5_metrics[4],"</td><td>",'<a href="PE_GE_poly5.pdf" target="_blank">Link</a>',"</td></tr>", "<tr><td>Polynomial regression with degree 6</td><td>",regmodel_poly6_metrics[1],"</td><td>",regmodel_poly6_metrics[2],"</td><td>",regmodel_poly6_metrics[3],"</td><td>",regmodel_poly6_metrics[4],"</td><td>",'<a href="PE_GE_poly6.pdf" target="_blank">Link</a>',"</td></tr>", "<tr><td>Generalized additive models</td><td>",regmodel_gam_metrics[1],"</td><td>",regmodel_gam_metrics[2],"</td><td>",regmodel_gam_metrics[3],"</td><td>",regmodel_gam_metrics[4],"</td><td>",'<a href="PE_GE_gam.pdf" target="_blank">Link</a>',"</td></tr>", "</table>", file = htmloutfile, append = TRUE); # Warning On # options(warn = oldw)