**0**wrote:

I have 74 samples with log normalised gene counts. I want to see their expression pattern changes with age (range 2-72 years). I am not binning the ages and using them as a continuous variable.

This is the script I used to get p-values using splines:

require(splines)

x<-ns(pheno_UMary$Age, df=5)

design<-model.matrix(~x)

fit<-lmFit(counts_UMary, design = design)

fit<- eBayes(fit, robust = TRUE)

results<-topTable(fit coef = 2:ncol(design), n=Inf, sort.by = "none")

head(View(results))

I am a little confused about how to interpret the x1-5 coefficients obtained from the splines.

I then run the second model is a simpler model using age directly as a covariate and use the logFC from that. Since this model assumes linearity, I will consider the adjusted p-values from the first model.

Also, even though we "sort.by = none" for both, the order is different so it is quite difficult to compare the results. And the p-values are extremely different as well.

**23k**• written 9 months ago by anupriya.dalmia •

**0**