search

gosianow / DRIMSeq_before_BioC3.6

Package for differential analyses using the Dirichlet Multinomial distribution
3 stars 0 forks source link

splice junction usage #5

Closed frankbioinfo closed 6 years ago

frankbioinfo commented 6 years ago

hello developer, thank you for this tool. this is not an issue but request your help.

Can this model be used to detect tissue specific splice junction usage across n=5 tissues?

I have splice junction counts from star and tophat aligners for each tissue( some of the have 2 replicates and others 3replicates).

could you please help me with model design if its possible , appreciate your help.

Many thanks.

kvittingseerup commented 6 years ago

Hi Frank

I think you are misunderstanding the exact purpose of DRIMSeq - it does differential usage analysis of an entire isoform - not a specific splice junction. This is an approach that computational is a bit harder but the biological interpretation is easier because you know the whole transcript.

If you are interested in specific exon usage I would suggest DEXseq as using all reads in the exons is more reliable than only considering junction spanning reads.

frankbioinfo commented 6 years ago

Hi @kvittingseerup

Thank you for reply. I did use DEXseq, DEXseq lacks (contrast) function where i can compare it across all the tissues, hence was wondering if I can use DRIMSeq.

kvittingseerup commented 6 years ago

Are you sure? From section 4 of the vignette it looks like DEXSeq can handle complex models.

If I was you I would ask the question on support.bioconductor.org and also describe what you have tried.