Hey All

I have analysed some RNA-Seq data for a mouse genome with both Cufflinks/Cuffdiff version 2.0 and 2.1 and get quite different results. I understand that there has been a major change for Cuffdiff between versions but the number of significant results identified in version 2.1 is quite striking, e.g version 2.0: 17 statistically (q) differential genes and version 2.1: 171 statistically (q) differential genes.

I was wondering if this level of difference was expected and/or if anybody else has experienced similar differences between Cuffdiff versions?

(I have posted a similar message on the tuxedo-tools Google group as well)

Kind regards


Additional info:

All genes significant (q) identified in version 2.0 were identified in 2.1.

The study is comparing wild type mice to a known ENU induced mutant. 4 replicates were used for both wild type and mutant.

Significantly expressed (q):
v2.0 -> genes: 17 isoforms: 0 splicing: 90
v2.1 -> genes: 171 isoforms: 78 splicing: 12

Diff expressed genes (p adjusted) using edgeR: 22
Diff expressed genes (p adjusted) using DESeq: 1

Commands used when comparing Cuffdiff versions:

/cufflinks-2.1.1.Linux_x86_64/cuffdiff -o /NGS/users/Thomas /Cufflinks_2.1.1/ -L WT,MUT -p 15 -c 0 --library-type=fr-unstranded <files>

/cufflinks-2.0.2.Linux_x86_64/cuffdiff -o /NGS/users/Thomas/Cufflinks_2.0.2/ -L WT,MUT -p 15 -c 0 --library-type=fr-unstranded <files>