Detecting GC Bias in Transcriptomes (RNA-seq)

Question

[Jon17]

What's the best way to detect GC bias in a transcriptome (RNA-seq)?

We are currently aligning to a regular genome fasta file (like hg38) but that means there are skips in the coverage. exon -- intron -- exon..... I know how to calculate GC bias of individual exons. Bedtools and R is a huge help there. But I'm stumped on how to do it for entire transcript. Any ideas?

I know how to do GC bias with Whole genome (CollectGcBiasMetrics Picard) and Sequence capture (custom scripting). But this transcript stuff is a bit more complex.

[ATpoint]

What is the final goal? Differential analysis? Check the alpine package at Bioconductor which implements GC bias correction. salmon implements this method in its --gcBias flag if you choose it to quantify your reads agsinst a transcriptome.

[Kristoffer Vitting-Seerup]

Appart from ATpoint's points (sorry for the pun - could not resist) you can also take a look at the R pakage cqn which have methods for identifying and correcting genome-wide GC-bias.

[mark.ziemann]

Would it be informative to calculate the GC content of the reads themselves?