Hi!

I want to use spliceR for detecting alternative splicing events and BitSeq to estimate differentially expressed isoforms. Is it possible to use the output of BitSeq for spliceR, to keep consistency in my analysis?

Thanks!

Hi,

after playing a bit with spliceR, I think that you have to use the function GRanges, see page 5 at: http://www.bioconductor.org/packages/release/bioc/vignettes/spliceR/inst/doc/spliceR.pdf

The input values corresponding to `gene_value` and `iso_value` should be gene and isoform expression levels respectively, as returned from `estimateExpression` and `getGeneExpression` commands of BitSeq.

Moreover, the input of `iso_p_value` should correspond to 2*(0.5 - |0.5 - pplr|), where pplr denotes the Probability of Positive Log Ratio as returned by `estimateDE` command of BitSeq (so that  values close to 0 indicate DE transcripts).

However, BitSeq output is not equivalent to Cufflinks, since no `gene_p_value` is currently reported from BitSeq. In case that this field is not necessary for the pipeline of spliceR then you can try the subsequent steps with BitSeq output as previously described.

Thank you very much!

I still have to run BitSeq, I wanted to use just the basic pipeline (getExpression and then getDE), is the output of getExpression equivalent to the one of estimateExpression you were referring to? And how can I specify the "iso-p-value"? I couldn't find it in the manual. Again, if I uderstood correctly, I can run the getGeneExpression after the basic pipeline, without running all the single steps, right?

I have other questions. The command for getGeneExpression requires the trInfoFile and a file like "data-GE.rpkm". The trInfoFile are produced during the parseAlignment step (the very first step), I imagine that if I run the basic pipeline I won't be able to retrieve these files in the future, so if I want to obtain the getGeneExpression I'll have to run BitSeq in the advanced way, or is there a way to store those trInfoFile after running the basic pipeline? I don't find however how the "data-GE.rpkm" files are obtained (those files are different from the "data.rpkm" file obtained with estimateExpression).

If someone has experience with that or have any suggestion I'd be very thankful!

Hi again,

Since you are planning to retrieve the temporary files in the future I would suggest to run BitSeq step by step, that is, `parseAlignment`, `estimateExpression`, `getMeanVariance`, `estimateHyperPar` and  `estimateDE`. The function 'getGeneExpression` should be used for gene estimates, given the output of `estimateExpression` and the *.tr file (which should contain both transcript and gene names).

Regarding your question about `iso-p-value`: this refers to the input of the GRanges function of spliceR package, so you will not find this terminology in BitSeq. As I already described, this should correspond to 2*(0.5 - |0.5 - pplr|), where pplr is reported in the output of `estimateDE` function of BitSeq.

Thank you very much!