dataframe 的数据广播和重新排序

Question

I have the following dataframe:

MOUSE_GENOTYPE  GENE_ID INFLAMMATION_TYPE   Fold gene expression

WT  Arg1                  microglia        0.02581
WT Arg1                  microglia        0.06783
KO  Arg1                  microglia        0.01477
KO Arg1               microglia        0.01787            
WT  Aspg                  PAN-reactive     0.01856
WT Aspg               PAN-reactive     0.08373
KO  Aspg                  PAN-reactive     0.0199
KO Aspg               PAN-reactive     0.09839
WT  Emp1                  A2-specific      0.03525
WT Emp1               A2-specific      0.02738
KO  Emp1                  A2-specific      0.01627
KO Emp1               A2-specific      0.02832

I'd like to reshape it by aggregating the mean and like the following:

                Arg1        Aspg                Emp1
MOUSE_GENOTYPE  microglia   PAN-reactive        A2-specific
WT              0.02581      0.0185             0.00691
KO              0.01477      0.0199             0.00631

I'd like to re-order the columns by the variable INFLAMMATION_TYPE so I can get PAN-reactive columns first, then microglia and then A2-specific.

                Aspg         Arg1                Emp1
MOUSE_GENOTYPE  PAN-reactive microglia          A2-specific
WT               0.0185      0.02581            0.00691
KO               0.0199      0.0147             0.00631

I have tried the dcast function:

results_reshaped <- dcast(results, 
                          MOUSE_GENOTYPE  ~ GENE_ID + INFLAMMATION_TYPE,
                          fun.aggregate = mean)

but the GENE_ID and INFLAMMATION TYPE get combined in one variable:

                         Arg1_ microglia     Aspg_ PAN-reactive            Emp1_ A2-specific
MOUSE_GENOTYPE  
WT                              0.02581  0.01856                                      0.00691
KO                              0.01477  0.0199                                      0.00631

I would like to keep them in separate columns and re-order them to get this. Suggestions?

Thanks! P

Answer 1

You can format names of your dataframe as next in order to reach a similar structure as you want:

#Format names
vecoriginal <- names(results_reshaped)
vecnames <- names(results_reshaped)[-1]
v1 <- c("",gsub("_.*", "",vecnames))
v2 <- c(vecoriginal[1],gsub(".*_","",vecoriginal[-1]))
#Assign names
names(results_reshaped)<-v1
#Create empty dataframe
empty <- as.data.frame(t(data.frame(v1,v2,stringsAsFactors = F)),stringsAsFactors = F)
names(empty)<-names(results_reshaped)
#Bind elements
DF <- rbind(empty,results_reshaped)
DF <- DF[-1,]
rownames(DF)<-NULL