RNASeq workflow inspired from Trapnell et al, 2012
From BITS wiki
The 2012 Nature Protocol article by Trapnell C and colleagues ([PMID: 22383036]) is very well written and introduces to the field of RNASeq with a nice full workflow. You will find below text and pieces of code that are directly derived from the original workflow and propose alternatives or discussions based on it.
# group reads and run tophat
* copy all read files in a common folder * rename each sample group with a unique prefix followed by '_reads.fq' (.gz compression is allowed) * edit the top of 'tophat_all.sh' to reflect your reference genome and transcriptome as well as the number of threads (cpu) avaiblable on your machine. * come to the current path in terminal # execute 'tophat_all.sh' tophat_all.sh
# run each pairwise cuffdiff comparison.
* move back to where the read files are stored * select two sample groups to be compared * execute 'cuffdiff_2.sh' and provide information to designate the two groups # example to compare NI and DEX groups using 16 threads cuffdiff_2.sh 16 "NI_" "DEX_"
# run a 4-way cuffdiff comparison.
* move back to where the read files are stored * select two sample groups to be compared * execute 'cuffdiff_4.sh' and provide information to designate the two groups # example to compare NI, DEX, GW, and DG groups using 16 threads cuffdiff_4.sh 16 "NI_" "DEX_" "GW_" "DG"
# run all four pairwise and the 4-way comparison in one go.
* move back to where the read files are stored * edit the bash script 'cuffdiffAll.sh' to match all sample groups and set the number of threads in it * execute 'cuffdiffAll.sh' cuffdiffAll.sh