Mercurial > repos > davidvanzessen > shm_csr
changeset 43:77a7ac76c7b9 draft
Uploaded
| author | davidvanzessen |
|---|---|
| date | Tue, 11 Apr 2017 08:02:17 -0400 |
| parents | 1cf60ae234b4 |
| children | 716622f5bb4f |
| files | plot_pdf.r shm_csr.py shm_csr.r wrapper.sh |
| diffstat | 4 files changed, 38 insertions(+), 10 deletions(-) [+] |
line wrap: on
line diff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/plot_pdf.r Tue Apr 11 08:02:17 2017 -0400 @@ -0,0 +1,17 @@ +library(ggplot2) + +args <- commandArgs(trailingOnly = TRUE) +print(args) + +input = args[1] +outputdir = args[2] +setwd(outputdir) + +load(input) + +print(names(pdfplots)) + +for(n in names(pdfplots)){ + print(paste("n:", n)) + ggsave(pdfplots[[n]], file=n, device="pdf") +}
--- a/shm_csr.py Tue Mar 28 08:25:36 2017 -0400 +++ b/shm_csr.py Tue Apr 11 08:02:17 2017 -0400 @@ -80,7 +80,7 @@ IDlist += [ID] -print mutationList, linecount +#print mutationList, linecount AALength = (int(max(mutationList, key=lambda i: int(i[4]) if i[4] else 0)[4]) + 1) # [4] is the position of the AA mutation, None if silent if AALength < 60:
--- a/shm_csr.r Tue Mar 28 08:25:36 2017 -0400 +++ b/shm_csr.r Tue Apr 11 08:02:17 2017 -0400 @@ -124,6 +124,8 @@ regions = c("CDR2", "FR3") } +pdfplots = list() #save() this later to create the pdf plots in another script (maybe avoids the "address (nil), cause memory not mapped") + sum_by_row = function(x, columns) { sum(as.numeric(x[columns]), na.rm=T) } print("aggregating data into new columns") @@ -308,7 +310,7 @@ print(p) dev.off() - ggsave(paste("transitions_stacked_", name, ".pdf", sep="")) + pdfplots[[paste("transitions_stacked_", name, ".pdf", sep="")]] <<- p png(filename=paste("transitions_heatmap_", name, ".png", sep="")) p = ggplot(transition2, aes(factor(reorder(variable, -order.y)), factor(reorder(id, -order.x)))) + geom_tile(aes(fill = value)) + scale_fill_gradient(low="white", high="steelblue") #heatmap @@ -316,7 +318,7 @@ print(p) dev.off() - ggsave(paste("transitions_heatmap_", name, ".pdf", sep="")) + pdfplots[[paste("transitions_heatmap_", name, ".pdf", sep="")]] <<- p } else { #print("No data to plot") } @@ -400,7 +402,7 @@ print(pc) dev.off() - ggsave("IGA.pdf", pc) + pdfplots[["IGA.pdf"]] <- pc } print("Plotting IGG piechart") @@ -423,7 +425,7 @@ print(pc) dev.off() - ggsave("IGG.pdf", pc) + pdfplots[["IGG.pdf"]] <- pc } print("Plotting scatterplot") @@ -445,7 +447,7 @@ print(p) dev.off() -ggsave("scatter.pdf", p) +pdfplots[["scatter.pdf"]] <- p write.table(dat[,c("Sequence.ID", "best_match", "VRegionMutations", "VRegionNucleotides", "percentage_mutations")], "scatter.txt", sep="\t",quote=F,row.names=F,col.names=T) @@ -471,7 +473,9 @@ print(p) dev.off() -ggsave("frequency_ranges.pdf", p) +pdfplots[["frequency_ranges.pdf"]] <- p + +save(pdfplots, file="pdfplots.RData") frequency_bins_data_by_class = frequency_bins_data
--- a/wrapper.sh Tue Mar 28 08:25:36 2017 -0400 +++ b/wrapper.sh Tue Apr 11 08:02:17 2017 -0400 @@ -176,6 +176,13 @@ echo "R mutation analysis" Rscript $dir/shm_csr.r $outdir/merged.txt $classes $outdir ${empty_region_filter} 2>&1 +echo "---------------- plot_pdfs.r ----------------" +echo "---------------- plot_pdfs.r ----------------<br />" >> $log + +echo "Rscript $dir/shm_csr.r $outdir/pdfplots.RData $outdir 2>&1" + +Rscript $dir/plot_pdf.r "$outdir/pdfplots.RData" "$outdir" 2>&1 + echo "---------------- shm_csr.py ----------------" echo "---------------- shm_csr.py ----------------<br />" >> $log @@ -249,7 +256,7 @@ echo "---------------- pattern_plots.r ----------------" echo "---------------- pattern_plots.r ----------------<br />" >> $log - Rscript $dir/pattern_plots.r $outdir/data_${func}.txt $outdir/aid_motives $outdir/relative_mutations $outdir/abolute_mutations $outdir/shm_overview.txt 2>&1 + Rscript $dir/pattern_plots.r $outdir/data_${func}.txt $outdir/aid_motives $outdir/relative_mutations $outdir/absolute_mutations $outdir/shm_overview.txt 2>&1 echo "<table class='pure-table pure-table-striped'>" >> $output echo "<thead><tr><th>info</th>" >> $output @@ -300,7 +307,7 @@ echo "<a href='aid_motives.pdf'><img src='aid_motives.png' /></a><br />" >> $output echo "<a href='relative_mutations.pdf'><img src='relative_mutations.png' /></a><br />" >> $output -echo "<a href='abolute_mutations.pdf'><img src='abolute_mutations.png' /></a><br />" >> $output +echo "<a href='absolute_mutations.pdf'><img src='absolute_mutations.png' /></a><br />" >> $output echo "<br />" >> $output cat $dir/shm_overview.htm >> $output echo "</div>" >> $output #SHM overview tab end @@ -657,7 +664,7 @@ echo "<tr><td>Base count for every sequence</td><td><a href='base_overview.html'>View</a></td></tr>" >> $output echo "<tr><td>The data used to generate the percentage of mutations in AID and pol eta motives plot</td><td><a href='aid_motives.txt' download='aid_motives.txt' >Download</a></td></tr>" >> $output echo "<tr><td>The data used to generate the relative mutation patterns plot</td><td><a href='relative_mutations.txt' download='relative_mutations.txt' >Download</a></td></tr>" >> $output -echo "<tr><td>The data used to generate the absolute mutation patterns plot</td><td><a href='absolute_mutations.txt' download='abolute_mutations.txt' >Download</a></td></tr>" >> $output +echo "<tr><td>The data used to generate the absolute mutation patterns plot</td><td><a href='absolute_mutations.txt' download='absolute_mutations.txt' >Download</a></td></tr>" >> $output echo "<tr><td colspan='2' style='background-color:#E0E0E0;'>SHM Frequency</td></tr>" >> $output echo "<tr><td>The data generate the frequency scatter plot</td><td><a href='scatter.txt' download='scatter.txt' >Download</a></td></tr>" >> $output