Potential inconsistency in buildSignatureMatrixUsingSeurat

[antparleo]

Hi, my name is Antonio. First of all, thank you for your time and feedback.

I am using your package and analyzing the code to understand how the function works. However, in the function of buildSignatureMatrixUsingSeurat I have identified a potential inconsistency. In comments, you mention that the genes are ordered by decreasing fold change, but in the actual code, they seem to be ordered by adjusted p-value. Please see your code:

#need to reduce number of genes
  #for each subset, order significant genes by decreasing fold change, choose between 50 and 200 genes
  #choose matrix with lowest condition number
  conditionNumbers<-c()
  for(G in 50:200){
    Genes<-c()
    j=1
    for (i in unique(id)){
      if(numberofGenes[j]>0){
        temp<-paste("cluster_lrTest.table.",i,sep="")
        temp<-as.name(temp)
        temp<-eval(parse(text = temp))
        temp<-temp[order(temp$p_val_adj,decreasing=TRUE),]
        Genes<-c(Genes,(rownames(temp)[1:min(G,numberofGenes[j])]))
      }

I am wondering whether the genes should be ordered by fold change or adjusted p-value. From my point of view, I consider that it would be more appropriate to order them by the fold change, but could you please confirm this?

Thank you so much for your help!!