The tutorial for a yale training session: TCGA RNA-seq Data, Download and Analyses all on your laptop.

See the slides used during the workshop here.

Go to TCGA data hub

• Navigate and select files to basket
• Download metadata and manifest from basket
• Download the files with GDC-client

Preprocess the metadata

• Convert to csv use the online tool json-to-csv
• Metadata Description here
• Choose and rename fields in a speadsheet or a R script.

.. Note: To to run the R script, you can install Rstudio.

Preprocess the FPKM matrix

• Convert the downloaded files to a FPKM matrix in unix shell/terminal

for f in */*.gz; do
  id=$(dirname $f); echo $id > $id.tmp; 
  zcat $f | cut -f2 >> $id.tmp; 
done
echo 'featureId' > tmp.index
zcat $f | cut -f1 >> tmp.index
paste tmp.index *.tmp > ../geneId_fileId_FPKM.txt
rm tmp.index; rm *.tmp

.. Note: to use linux shell, run terminal on mac (OS X); install and run babun on a PC (windows).

• Description of the Barcode
• Description of the pipeline
• Download the GENCODE gene annotation file
• Map the FPKM matrix to gene symbol and barcode with preprocess_count_matrix.R.

Introduction of analyses in R

Using the script to:

• Filter the genes and convert FPKM to log scale
• Id genes coexpressed with your gene of interest
• Id genes differently expressed between paired normal and tumor
• PCA plot

Introduction of the analyses by FireHose

• Gene
• Cohort summary
• Cohort data and workflow
• Cohort analysis

FAQS

quick fix to get the miRNA FPM matrix for Queen Okoro

• Convert the downloaded files to a FPKM matrix in unix shell/terminal

# cd to the folder with all the txt files under each sample directory.
for f in */*.txt; do
  id=$(dirname $f); echo $id > $id.tmp; #colnames to-be
  cat $f | cut -f3 >> $id.tmp;  #cell values to-be
done
echo 'featureId' > tmp.index
cat $f | cut -f1 >> tmp.index #the rownames to be
paste tmp.index *.tmp > ../featureId_fileId_FPKM.txt
rm tmp.index; rm *.tmp