Hi,
I have just re-analysed my data to get TPM rather than FPKM values. I have done this using RSEM.
A google search has informed me that the statistics behind edgeR and DESeq are not compatible with the 'over the whole genome/transcriptome' approach of RSEM (simply put).
My question now is, whether anybody has tried using the bam output files produced by RSEM as input for cuffdiff2. Can you use the cuffdiff algorithm on TPM values and just "trick the system" or is there anything that suggests not to do so? Alternatively, what do you use for differential gene expression analysis after you have the RSEM output?
My complete suggested new workflow is:
raw data -> fastqc -> trimmoPE -> fastqc -> RSEM-calculate-expression and then potentially cuffdiff2 -> cummeRbund -> heatmaps, pathway analysis etc.
Thanks for your help!
I have just re-analysed my data to get TPM rather than FPKM values. I have done this using RSEM.
A google search has informed me that the statistics behind edgeR and DESeq are not compatible with the 'over the whole genome/transcriptome' approach of RSEM (simply put).
My question now is, whether anybody has tried using the bam output files produced by RSEM as input for cuffdiff2. Can you use the cuffdiff algorithm on TPM values and just "trick the system" or is there anything that suggests not to do so? Alternatively, what do you use for differential gene expression analysis after you have the RSEM output?
My complete suggested new workflow is:
raw data -> fastqc -> trimmoPE -> fastqc -> RSEM-calculate-expression and then potentially cuffdiff2 -> cummeRbund -> heatmaps, pathway analysis etc.
Thanks for your help!
Comment