Hi! There’s a significant difference in gene score between CERES (~-0.7) and Chronos (~-0.1), for my gene of interest in a large subset of cell lines. Is there a way to understand which algorithm is more representative in this case? (in vitro studies strongly suggest CERES is the better measure). All the best.
Hello Matthew,
There are many possible reasons for these scores to differ depending on which versions of the data you are looking at. CERES has not been run since I believe 2021, and was never run with more than one library, does not model screen bias, and is prone to pegging its estimates to strongest outlier guide in the set. That said, there will certainly be some cases where CERES’s estimates are more accurate. If you want to understand a specific gene’s viability effect in detail, you’ll want to look at the guide-level data which is included in every dataset.
Thanks Joshua - appreciate you replying with this suggestion.
Is there a link to information on how I can download the guide level data, or are you able to briefly explain how I can obtain this information? Thanks again and best regards, Matt
Hi Matt, files ending in Readcounts or LogfoldChange are guide-level. You can find more information in the README for each dataset. Take a look at the portal’s Downloads page, and let me know if you have any questions.
