Post-Cufflinks analysis of a single transcriptome, no comparison

[sindrle]

Hi!
Say you have a new cell type thats never been RNA sequenced, but from a known species. How you go about it after Tophat2->Cufflinks?

I started out sorting by FPKM just to see whats high/low/not expressed.

Now I want to facilitate further analysis for my supervisor:

a. Using bioDB.net, finding protein/non-protein coding genes
b. GO terms for each gene
c. Ideas?

Problem is that bioDB.net has an input limit far below the number of genes I have...

Also, is there a way of making each gene symbol into a hyperlink (in Excel) directly to a search in PubMED gene for this gene? Or equal/better solution?


Thank you so much!

[gringer]

Cufflinks produces GTF files for the discovered transcripts, which can be used to generate a transcriptome using the reference genome. There are various programs for discovering protein sequences and annotating de-novo transcriptomes (e.g. FrameDP or Blast2GO), and for lifting annotations from other transcriptomes (e.g. RATT).

[sindrle]

Ok, thanks!

But its from mus musculus, only single-cell not charachterized yet.

[gringer]

Do you have reason to believe that the transcript annotation for a single cell from Mmus will differ from that of the reference Mmus transcriptome? You might get different transcripts expressed, but I would find it odd if the annotation were different.

[sindrle]

I have no reason to believe the annotation is different.

I ran Cufflinks with mm10 GTF and with novel discovery.

A also compared to two other single cell types: Nerve and Epithelium, using Cuffdiff.

Question is what I do from here?

Sorry, its my first time working with "just see whats going on in this cell" type of study.

[gringer]

Ah, right. Something that people have asked me to do following on from this in a similar situation is some form of enrichment analysis:

http://www.broadinstitute.org/gsea/msigdb/index.jsp [gene set enrichment]
http://www.bioconductor.org/packages...tml/goseq.html [GO enrichment]

Also loading tracks in a genome browser can be helpful, if there are particular genes of interest.

edit: there's also CummeRbund, which is the next step of the Tuxedo suit.

[sindrle]

Ok, thanks, but remember its not differential gene expression Im worried about now. Only learning about the gene expression in the cell.

Btw, do you know how to not use the "XLOC" ids for transcripts? They are non-informative to me.

I have used mm10 GTF as well as a GTF made by Cuffmerge.