Reproducing the PA values from the paper

[zslastman]

Hi guys - to double check some of my own work, I've been trying to compare it to scikit-ribo's estimates of predicted protein abundance (PA). If I read the paper and code correctly, then it should be possible to get the PA estimates by multiplying the RNAseq TPMs by the TE estimates that scikit-ribo outputs (?). Doing this with your example data however, and comparing it to the proteomics data form Lawless et al, gives me a (much) worse correlation with the proteomics than just plain-old riboseq TPMs from salmon. From the paper, it should be much better, say .83 vs .77 with TPMs alone.

Am I calculating PA values correctly? what exactly should I multiply scikit-ribo's output by to get them?