Hi I am trying to batch correct for samples from 4 different experiments that I ran in one fluidigm plate. I tried normalization with liima using

norm_exprs <- limma::normalizeBetweenArrays(exprs, method="quantile")

followed by

batch <- factor(data$Experiment)
# Perform ComBat batch correction
modcombat <- model.matrix(~ Group, data=data)
combat_data <- sva::ComBat(dat=norm_exprs, batch=batch, mod=modcombat, par.prior=TRUE, prior.plots=FALSE)

The output of my data looks completely changed, with many negative values, and fewer significant genes than what I was seeing before normalization and batch correction. One of my most significant genes before correction is now not significant, and when I checked the corrected data I saw that some values are changed such that the trend is completely opposite to that of my raw data.

Is there a better way to normalize and batch correct fluidigm data?

The batch correction should simply be included in the linear model by

design <- model.matrix(~ Group + Batch)

No need for ComBat. Using an external batch correction program like ComBat is unnecessary and will make the limma DE analysis results anti-conservative.

Of course batch correction may still change your results. If it didn't, there would be no need for batch correction in the first place.

Regarding normalization, the quantile normalization is probably ok but I wonder whether the Fluidigm data is being normalized to reference genes. Instead of quantile normalization, I usually use cyclic loess to normalize qPCR ct values to one or more reference genes.