What are the recommended sequencing depth and length for conducting analyses on alternative splicing?

[Jon Gleur]

I am aware that 20 million reads are considered sufficient for conducting differential expression analyses. Is there a specific cut-off or recommended number of reads for alternative splicing analyses? Additionally, would 150 bp paired-end sequencing be significantly better than 100 bp paired-end sequencing?

[kutsc...@gmail.com]

http://dx.doi.org/10.1073/pnas.1419161111
There's a section "The Influence of Sample Size and Sequencing Depth on Detection Accuracy" with some results. The lowest number of reads in a replicate discussed there is 20M reads (200M reads over 10 replicates)

https://doi.org/10.1093/bib/bbz126
> Our analysis reconfirmed the DS results to be quite robust after 40 to 60 million reads per sample

https://knowledge.illumina.com/library-preparation/rna-library-prep/library-preparation-rna-library-prep-reference_material-list/000001243
> * Experiments looking for [...] some information on alternative splicing, typically require 30 million to 60 million reads per sample
> * Experiments looking to get an in-depth view of the transcriptome, or to assemble new transcripts, may require 100 million to 200 million reads
> Novel transcriptome assembly and annotation projects tend to benefit from longer, paired-end reads (such as 2 x 75 bp or 2 x 100 bp) to enable more complete coverage of the transcripts and identification of novel variants or splice sites

rMATS only needs the read to cover the splice junction. I think for rMATS the benefit of longer reads is that it's more likely for a read to be uniquely mapped

Eric