CUfflink data for statistical test

Question

[1769mkc]

Can I use normal statistical test on cufflink data since its already normalised , so can i use parametric statistics on those normalised data and find differential expression , instead of running doing a cuff-diff analysis ?

Any suggestion would be highly appreciated

Answer 1

[andrew.j.skelton73]

If you want to use a transcriptome assembler, I think you'd be much safer using Stringtie, as Tophat2 + Cufflinks are in low support / deprecated. There's a method for making these results compatible with DESeq2 / edgeR, see the section "Using StringTie with DESeq2 and edgeR" on that link.

[1769mkc]

well i already did the tophat protocol but Im looking to test differential expression apart from cuffdiff , like can i apply T- test on the cufflink nopmralised data ?

[grant.hovhannisyan]

FPKM/RPKMs are not directly comparable between samples. See this video of Lior Pachter https://youtu.be/5NiFibnbE8o?t=1833

In general to get interpretable and trustworthy results, it is almost always better to use generally accepted piece of software. In this case you also can have feedback from community. For your case, have a look here http://bioconductor.org/packages/release/bioc/html/tximport.html