differential gene expression analysis between Different strains by RNA-seq

[qqtwee]

Hello,
I work with solexa paired-end RNA-seq data of bacteria. I analyze differential gene expression by FPKM from cufflinks,and analyze FPKM using DEGseq to get differential expression genes. I have different strains of bacteria, and their genomes are pretty similar, but not identical. that means, I get FPKM from cufflinks for different strains using different reference and different annotation files(that is different gtf files), then can I directly compare FPKM, if not, how can I normalize FPKM to analyze different strain?
I am looking forward to your reply. Thank you!

[Simon Anders]

Look through older threads here to learn why you should not use FPKM values anyway for differential expression testing.

[DZhang]

Hi qqtwee,

I am not sure this type of analysis is regular in your research or not. There are quite a few standard options in analyzing differential gene expression (e.g., count-base or FPKM-based) and in data normalization. You do, however, have a unique situation as the genomes are different. What I recommend, if this is something you will be doing regularly in your research, is to analyze these data in all the combinations to see how sensitive the outcome is relating to quantitation methods, normalization methods, and (more interesting to me) levels of genome difference. Then you try to experimentally validate these results to see which analysis approach works best.

I know this is lots of work. But if you want to get the right results, this is what it takes, IMHO.

Douglas

https://www.contigexpress.com

[mbblack]

Hello,
I work with solexa paired-end RNA-seq data of bacteria. I analyze differential gene expression by FPKM from cufflinks,and analyze FPKM using DEGseq to get differential expression genes. I have different strains of bacteria, and their genomes are pretty similar, but not identical. that means, I get FPKM from cufflinks for different strains using different reference and different annotation files(that is different gtf files), then can I directly compare FPKM, if not, how can I normalize FPKM to analyze different strain?
I am looking forward to your reply. Thank you!

As long as each strain has it's own control group, you would just analyze each strain independently for differentially expressed genes. Then, as long as you used annotation that includes some element to align the data sets by homology (entrez gene ID, refseq gene ID - something that allows you to align homologous genes), you would then make your cross-species comparisons for those homologous genes in common.

So, just determine your strain specific differentially expressed genes, and do your cross species/strain comparison with those differentially expressed genes which are also homologous between the two strains.