Using "combn" to create list of dataframes for all combinations of a variable

Question

Using my dataframe below and working in R, I would like to generate a list of new dataframes in which each dataframe is based on a unique combination of 4 different values of my Taxon vector. In the dataframe below I have 10 different Taxon (many have multiple replicate entries) and I would like to select 4 Taxon at a time, for all combinations of 4. I believe that there should be 210 different combinations of 4 Taxon from these 10 Taxon (when sampled without replacement and order is not important). So ultimately I would like a list containing 210 dataframes, each containing a different combination of 4 Taxon (with all replicate rows for each Taxon)!!!

Though selection is based on the Taxon vector, I would like the new dataframes to include the other columns of information. For instance, if "Aphididae.sp.3" is selected, i would like the new dataframe to also have -25.92(C), 1.69(N), sap(func.group), herbivore(trophic.group) listed in columns alongside.

So far I used the "combn" function to generate all combinations of 4 taxon using the "unique" command, but I can't get all the other columns of information included in this, and it does not give all the replicate entries for each taxon! I'd be grateful for any help!

my code:

combos<-combn(unique(df$Taxon),4)

df:

                  Taxon      C    N     func.group trophic.grp
1           Aphididae.sp.3 -25.92 1.69        sap   herbivore
2           Aphididae.sp.3 -25.91 1.78        sap   herbivore
3           Aphididae.sp.3 -26.05 1.74        sap   herbivore
4  Cicadellidae.mixed.juvs -28.94 1.19        sap   herbivore
5  Cicadellidae.mixed.juvs -29.25 2.24        sap   herbivore
6  Cicadellidae.mixed.juvs -28.17 1.88        sap   herbivore
7  Cicadellidae.mixed.juvs -28.29 1.94        sap   herbivore
8        Cicadellidae.sp.1 -27.69 2.25        sap   herbivore
9        Cicadellidae.sp.1 -27.67 2.41        sap   herbivore
10       Cicadellidae.sp.1 -26.65 3.26        sap   herbivore
11       Cicadellidae.sp.1 -28.30 3.20        sap   herbivore
12       Cicadellidae.sp.1 -28.08 1.88        sap   herbivore
13       Cicadellidae.sp.2 -26.59 2.89        sap   herbivore
14       Cicadellidae.sp.3 -26.82 5.16        sap   herbivore
15       Cicadellidae.sp.4 -26.54 3.46        sap   herbivore
16       Cicadellidae.sp.4 -26.55 4.05        sap   herbivore
17       Cicadellidae.sp.4 -27.20 3.14        sap   herbivore
18       Cicadellidae.sp.4 -26.48 3.80        sap   herbivore
19       Cicadellidae.sp.5 -27.54 4.17        sap   herbivore
20       Cicadellidae.sp.5 -27.18 3.43        sap   herbivore
21       Cicadellidae.sp.5 -27.46 4.03        sap   herbivore
22       Cicadellidae.sp.6 -26.71 1.09        sap   herbivore
23       Cicadellidae.sp.6 -26.33 1.56        sap   herbivore
24       Cicadellidae.sp.6 -25.59 0.59        sap   herbivore
25       Cicadellidae.sp.6 -25.07 0.84        sap   herbivore
26       Cicadellidae.sp.6 -26.56 0.97        sap   herbivore
27       Cicadellidae.sp.7 -25.84 1.08        sap   herbivore
28       Cicadellidae.sp.7 -24.96 1.36        sap   herbivore
29       Cicadellidae.sp.7 -26.15 1.90        sap   herbivore
30       Cicadellidae.sp.7 -26.58 2.63        sap   herbivore
31       Cicadellidae.sp.8 -28.02 2.28        sap   herbivore
32       Cicadellidae.sp.8 -27.90 2.01        sap   herbivore
33       Cicadellidae.sp.8 -27.70 1.92        sap   herbivore
34       Cicadellidae.sp.8 -26.85 1.04        sap   herbivore

Answer 1

combos should be a 4 x 210 matrix of character vectors. If you instead of your original code use:

 combos<-combn(unique(as.character(df$Taxon)), 4)   # factors would be converted

You should be able to pass those vectors to a subsetting function with apply:

combdf <- apply(combos, 2, function(vec) df[ df$Taxon %in% vec, ] )

During testing I discovered that the original matrix got messed up because I had left the Taxons in as factors, hence the need for the as.character call before unique. I didn't see this until attempting the apply call since I got 210 items but most of them were empty.

Answer 2

lapply(ncol(combos), function(x) df[df$Taxon %in% combos[,x],])