This project describes how to generate the example data for DEIVA, https://github.com/Hypercubed/DEIVA.

The example data are from this publication: [1] Kratz, Anton, et al. "Digital expression profiling of the compartmentalized translatome of Purkinje neurons." Genome research 24.8 (2014): 1396-1410. http://genome.cshlp.org/content/24/8/1396

Four differential expression tests are performed: bound vs unbound, and bound membrane vs unbound membrane, with DESeq2 and edgeR, respectively.

Get the input data from [1]:

wget http://genome.cshlp.org/content/suppl/2014/06/05/gr.164095.113.DC1/Supplemental_Table.S3.xlsx

Export the first sheet as ASCII. To do this open the table with a spreadsheet application, here I use gnumeric 1.12.24.

"Data -> Export Data -> Export as CSV file..."

Now I have the data in a file Supplemental_Table.S3.csv

Simplify the file name, remove columns not relevant in this context, replace the comma with TAB:

mv Supplemental_Table.S3.csv S3.csv
sed 's/,/\t/g' S3.csv | cut -f 1-5,10-27 > data/expr_table.bound_vs_unbound.csv
sed 's/,/\t/g' S3.csv | cut -f 1,2,4,10-23,26,27 > data/expr_table.bmemb_vs_bcyto.csv

Delete the id field with vi. the file should not start with a TAB (i.e. delete the TAB after "id").

I can now use expr_table.bound_vs_unbound.csv and expr_table.bmemb_vs_bcyto.csv as DESeq2 input.

Manually prepare a expr_table.desc.bound_vs_unbound.csv and expr_table.desc.bmemb_vs_bcyto.csv file.

Execute this in RStudio, or just run:

R CMD BATCH Rscripts/edgeR.bound_vs_unbound.R Rdump/edgeR.bound_vs_unbound.Rout
R CMD BATCH Rscripts/edgeR.bmemb_vs_bcyto.R Rdump/edgeR.bmemb_vs_bcyto.Rout

Add columns with gene symbol and representative cluster. The idea is to sort the original sheet, sort the results file, and add the two columns.

tail -n+2 out/edgeR/edgeR.bound_vs_unbound.txt | sort -k 1,1 > srtd.payload
tail -n+2 S3.csv | sort -k 1,1 | cut -d "," -f 1,64 | sed 's/,/\t/g' > srtd.S3.csv
paste srtd.payload srtd.S3.csv | cut -f 1-6,8 > foobar 
cat data/header_wh_symbol foobar >> annotated/edgeR.bound_vs_unbound.tsv

tail -n+2 out/edgeR/edgeR.bmemb_vs_bcyto.txt | sort -k 1,1 > srtd.payload
tail -n+2 S3.csv | sort -k 1,1 | cut -d "," -f 1,64 | sed 's/,/\t/g' > srtd.S3.csv
paste srtd.payload srtd.S3.csv | cut -f 1-6,8 > foobar 
cat data/header_wh_symbol foobar > annotated/edgeR.bmemb_vs_bcyto.tsv

rm srtd.payload
rm srtd.S3.csv
rm foobar

DESeq2IVA_bound_vs_unbound.tsv can now be loaded into DESeq2IVA, done.

Now also do this to generate the DESeq2-based input files. This uses the same input files, so I keep this in the same project.

R CMD BATCH Rscripts/DESeq2.bound_vs_unbound.R Rdump/DESeq2.bound_vs_unbound.Rout
R CMD BATCH Rscripts/DESeq2.bmemb_vs_bcyto.R Rdump/DESeq2.bmemb_vs_bcyto.Rout

Final results files w/o symbol: txt files in out.

Final results files w/h symbol: csv files in annotated.

tail -n+2 out/DESeq2/DESeq2.bound_vs_unbound.txt | sort -k 1,1 > srtd.payload
tail -n+2 S3.csv | sort -k 1,1 | cut -d "," -f 1,64 | sed 's/,/\t/g' > srtd.S3.csv
paste srtd.payload srtd.S3.csv | cut -f 1-7,9 > foobar 
cat data/d2.header_wh_symbol foobar >> annotated/DESeq2.bound_vs_unbound.tsv

tail -n+2 out/DESeq2/DESeq2.bmemb_vs_bcyto.txt | sort -k 1,1 > srtd.payload
tail -n+2 S3.csv | sort -k 1,1 | cut -d "," -f 1,64 | sed 's/,/\t/g' > srtd.S3.csv
paste srtd.payload srtd.S3.csv | cut -f 1-7,9 > foobar 
cat data/d2.header_wh_symbol foobar >> annotated/DESeq2.bmemb_vs_bcyto.tsv

rm srtd.payload
rm srtd.S3.csv
rm foobar