[Cuffdiff] Differentially Expressed Genes and Transcript Analyses

Question

[Donny]

Hi everyone.

My name is Donny, bioinformatician from Malaysia.

Currently I am using Cuffdiff, trying to explore the differentially expressed genes within a sets of data (without replicates). There are a total of 6 condition overall and I would like to seek for some advice here with the Cuffdiff command lines. Thank you very much for that! =)

I understand that if we were comparing 2 conditions, the command line would be something like:-

$ cuffdiff -o diff_out -b genome.fa -p 8 -L C1,C2 -u merged_asm/merged.gtf \
./C1_R1_thout/accepted_hits.bam,./C1_R2_thout/accepted_hits.bam,./C1_R3_thout/accepted_hits.bam \
./C2_R1_thout/accepted_hits.bam,./C2_R3_thout/accepted_hits.bam,./C2_R2_thout/accepted_hits.bam

This command line was based on the TopHat paper.

Note: C denotes for sample, while R denotes for replicates.

What if I have 6 conditions (without replicates)?

Should the command line be something like:-

$ cuffdiff -o diff_out -b genome.fa -p 8 -L C1,C2,C3,C4,C5,C6 -u merged_asm/merged.gtf \
./C1_thout/accepted_hits.bam, ./C2_thout/accepted_hits.bam, ./C3_thout/accepted_hits.bam \
./C4_thout/accepted_hits.bam, ./C5_thout/accepted_hits.bam, ./C6_thout/accepted_hits.bam

Later on, I hoped to be able to compare the groups like C1 vs C3 or C4 vs C6 etc..

Thank you very much in advance!

Answer 1

[Satyajeet Khare]

1. You can compare two conditions at a time. You can perform multiple of such one-on-one comparisons using contrast file. But all-versus-all is not possible to the best of my knowledge.
2. The old tuxedo protocol is replaced. The new tuxedo protocol (HiSAT, Stringtie, Ballgown) or other count matrix based methods (EdgeR) might be handy.
3. Normalisation will be a serious problem without replicates. Not sure how accurate the analysis will be.