This supplementary contains the especially the code to run mageck on our data, and is effectively the documentation for how to map
the original names (in hind-sight, crappily chosen!) to the final output files. Note that not all counts are used - we did some experiments
that were not included in the paper because we were not sure about the quality, or how to best integrate it. Yet the data is provided here as-is.

The output of mageck has been provided for convenience.

Some notes:

1. The list of hits is quite noisy. In the paper we filtered by gene expression to remove the worst false positives

2. We confused the samples Xbp+ and Xbp- (the label came off the tube in the last second...). In the paper we only compare genes by the sign (ignoring up or down) to circumvent this

3. We had to reamplify the library throughout the experiments, and sometimes we used antibodies, other times reporter mice. So for a given gene, we do NOT have perfect experimental repeats.
   This makes the statistical analysis really tricky. A plain MLE model as is used by Mageck2 will likely not do the trick. Our Baiopse algorithm includes a screen efficiency parameter
   to try and circument this.

4. In hind-sight, and looking at more recent CRISPR papers, a different statistical model might have been suitable. The transduction might have been a harsher stage for the complexity than
   we expected. This is now easier to track using CRISPR-UMI-based protocols

