Rrtk2 / RRtest

My personal package filled with tools to perform data analyis

GNU General Public License v3.0

0 stars 1 forks source link

Resulting standardized effect size #2

Closed Rrtk2 closed 1 year ago

Rrtk2 commented 4 years ago

Can we have a version of KDEA that does this which reports standardized effect size instead of logFC like this ?temp_fit$coefficients / temp_fit$stdev.unscaled 2:23 when the outcome variable is not binary but contnious?

Rrtk2 commented 1 year ago

Closed due to aging.

However first part can be extracted now (a is output from KDEA):

"Effect" metric

get effect Cohen's D metric

S2Post_median = apply(do.call(cbind,lapply(a$RawLimmaRes,function(x){x$s2post})),1,median) Cohens_d = a$Rankobject$MedianLogFC / sqrt(S2Post_median[a$Rankobject$FeatureName])

Continuous KDEA future