Chronos postprocessing

I have two questions regarding the late phases of Chronos algorithm.

  1. After estimating the fractional change in growth rate for gene g in cell line c,

r_{c,g} = (R*_{c,g} / R_c) - 1

is there a scaling (such has forcing the median of known essential (resp. non-essential) genes to -1 (resp. 0)) or r_{c,g} is the actual quantity found in Achilles_gene_effect_Chronos.csv file?

  1. Are the probabilities in Achilles_gene_dependency_Chronos.csv computed as in CERES, that is the probability of the score being generated by the distribution of essential gene scores? If yes, it is done with BAGEL algorithm, or with a from-scratch EM, or with other?

Thanks for your work, and this forum!


for 1., there is a scaling applied. The whole matrix is multiplied by a constant such that the median of all common essentials is around -1. For 2., it is done as in CERES, with a different but similar EM method to BAGEL.

1 Like

Great, thanks a lot! :grin: