Cufflinks App Workflow
| 1 | Novel transcript assembly. If novel transcript assembly is requested, transcripts are first assembled and quantified independently for each sample. Cuffmerge merges the assemblies between the samples of each group and combines them with the known gene models from annotation. Cuffcompare then provides annotations of novel transcripts, for example, if they overlap with a known transcript. Lastly abundances will be re-estimated for each transcript, known or novel. |
| 2 | Differential expression. Cuffdiff calculates differential expression between these two sample groups, estimating variance from the individual samples supplied for each sample group. Cuffdiff calculates a variety of information regarding differential expression, which are summarized in the on-line manual (cole-trapnell-lab.github.io/cufflinks/cuffdiff/index.html#cuffdiff-output-files) for that tool. |
Cuffdiff needs a consensus set of transcripts that is compared between different sample groups. If no novel transcript detection is performed, this consensus set is simply the reference annotation. If novel transcript assembly is requested, then the set of transcripts from each sample must first be merged together. The Cuffmerge tool does this, while also merging in the known gene models from annotation.
Figure 10 TopHat and Cufflinks Workflow
