effect sizes vs power

[coralzhang]

I am trying to understand the power calculation function power.general.withDoublets. I created 3 reference studies, with low, medium and high fold change values; I expect the power to increase with every setting being fixed. But the power provided was lowest for the medium fold changes. Can you explain why? ` dat=scPower::de.ref.study dat2=dat[dat$name=="Blueprint (CLL) uCLL-iCLL",] dat2=dat2[order(dat2$FoldChange),]

dat3=dat2[11:15,]; dat3$rank=10001:10005 dat4=dat2[121:125,]; dat4$rank=10001:10005 dat5=dat2[331:335,]; dat5$rank=10001:10005

power3<-power.general.withDoublets(nSamples=10,nCells=7000,readDepth=25000,ct.freq=0.3, type="de", ref.study=dat3, ref.study.name="Blueprint (CLL) uCLL-iCLL", samplesPerLane=2,read.umi.fit = scPower::read.umi.fit[read.umi.fit$type=="10X_PBMC_1",], gamma.mixed.fits = scPower::gamma.mixed.fits,ct="CD4 T cells", disp.fun.param=scPower::disp.fun.param,mappingEfficiency = 1, min.UMI.counts = 3,perc.indiv.expr = 0.9,sign.threshold = 0.05,MTmethod="Bonferroni", multipletRateGrowth="constant")

power4<-power.general.withDoublets(nSamples=10,nCells=7000,readDepth=25000,ct.freq=0.3, type="de", ref.study=dat4, ref.study.name="Blueprint (CLL) uCLL-iCLL", samplesPerLane=2,read.umi.fit = scPower::read.umi.fit[read.umi.fit$type=="10X_PBMC_1",], gamma.mixed.fits = scPower::gamma.mixed.fits,ct="CD4 T cells", disp.fun.param=scPower::disp.fun.param,mappingEfficiency = 1, min.UMI.counts = 3,perc.indiv.expr = 0.9,sign.threshold = 0.05,MTmethod="Bonferroni", multipletRateGrowth="constant")

power5<-power.general.withDoublets(nSamples=10,nCells=7000,readDepth=25000,ct.freq=0.3, type="de", ref.study=dat5, ref.study.name="Blueprint (CLL) uCLL-iCLL", samplesPerLane=2,read.umi.fit = scPower::read.umi.fit[read.umi.fit$type=="10X_PBMC_1",], gamma.mixed.fits = scPower::gamma.mixed.fits,ct="CD4 T cells", disp.fun.param=scPower::disp.fun.param,mappingEfficiency = 1, min.UMI.counts = 3,perc.indiv.expr = 0.9,sign.threshold = 0.05,MTmethod="Bonferroni", multipletRateGrowth="constant")

`

here is the output: power3: name powerDetect exp.probs power sampleSize 1 Blueprint (CLL) uCLL-iCLL 0.9889792 0.9900965 0.9988715 10 totalCells usableCells multipletFraction ctCells readDepth readDepthSinglet 1 7000 7000 7.67e-06 2100 25000 25000 mappedReadDepth expressedGenes 1 25000 11210

power4: name powerDetect exp.probs power sampleSize 1 Blueprint (CLL) uCLL-iCLL 0.501316 0.9900965 0.5063288 10 totalCells usableCells multipletFraction ctCells readDepth readDepthSinglet 1 7000 7000 7.67e-06 2100 25000 25000 mappedReadDepth expressedGenes 1 25000 11210

power5: name powerDetect exp.probs power sampleSize totalCells 1 Blueprint (CLL) uCLL-iCLL 0.9900965 0.9900965 1 10 7000 usableCells multipletFraction ctCells readDepth readDepthSinglet 1 7000 7.67e-06 2100 25000 25000 mappedReadDepth expressedGenes 1 25000 11210