I did a two-class comparison for the PRISM Repurposing Public 23Q2 data for a few lineages of interest and found that the distribution of effect sizes is not centering at 0 (whereas in the Skyros internal data portal it did). Seems like some systematic probably with calculating the effect size?
For Ewing lines:
For rhabdomyosarcoma lines:
For dedifferentiated/well-differentiated liposarcoma lines:
1 post - 1 participant