broadinstitute / lincs-profiling-complementarity

Analyzing and comparing signal found in different profiling technologies
BSD 3-Clause "New" or "Revised" License
5 stars 5 forks source link

[Response to Review] Coefficient of variation supplementary figure #55

Closed gwaybio closed 2 years ago

gwaybio commented 2 years ago

Adding Coefficient of Variation (CV) code to calculate CV of level 4 CP and L1000 features. I also add the code to generate the figure.

cv-supfig

Legend: Tracking feature variation in Cell Painting and L1000 profiles. Feature mean, standard deviation, and coefficient of variation (CV; standard deviation / mean) in (a) Cell Painting and (b) L1000 profiles. Note the different axis scales and common legend colors. Each point represents a feature, and the grey points have a mean between -0.3 and 0.3 to minimize extremely large CV values. (c) CV distributions for Cell Painting and L1000 profiles demonstrate higher relative variability in Cell Painting features. (d) Per replicate cv comparing replicate treatments for Cell Painting (x axis) and L1000 (y axis). Each point represents the log10 mean absolute value CV for all features measured within each replicate treatment. We remove treatments with CV values equal to zero. All panels display level 4 Cell Painting spherized profiles and level 4 L1000 nonspherized profiles.

This figure likely belongs after original Supplementary Figure 9.

gwaybio commented 2 years ago