Hi,
When one have generated counts using FeatureCounts it generates a list of transcripts and note genes. I understand the procotol as "gene"-counts should be used - however how is genecounts defined? Can I use the transcript-count-lists as input in the DESeq2 analysis?
Or should one convertet the transcript-counts (generated from featurecounts) to gene-counts? Is that what you are explaining in the protocol under: "Transcript abundance files and tximport input"? I dont really follow:
Concluding question: Is it fine to use transcript-counts from FeatureCounts as input?
Best regards,
S
thank you for your answer. I am still curious if I can use DESeq to to analys the differential expression of each transcript. Would there be a problem with using transcript as input insteed of genes? Given that I would like to look at each transcript seperatly and not the gene as a whole.
Thank you
S
We profiled this in a recent F1000Research publication. See the section "DTE evaluation".
DESeq2 on transcript counts seems to work decently (room for improvement, which we are working on), but an important caveat is that we quantified transcript expression using Salmon. I would recommend to use a transcript quantification method such as Salmon, kallisto, RSEM, etc. to obtain transcript counts. I collaborated on Salmon, and I think the advantage to using this method is that is pretty good at resolving between differences in coverage due to usage of specific isoform(s) and coverage variability due to technical biases.