Subset columns of one data frame according to another data frame's rows

Question

I would like to subset some of its columns according to another data frame's rows. So the two data frames are as shown below:

df1 <- structure(list(ID = structure(c(3L, 1L, 2L, 5L, 4L), .Label = c("cg08", "cg09", "cg29", "cg36", "cg65"), class = "factor"), chr = c(16L, 3L, 3L, 1L, 8L), gene = c(534L, 376L, 171L, 911L, 422L), GS12 = c(0.15, 0.87, 0.6, 0.1, 0.72), GS32 = c(0.44, 0.93, 0.92, 0.07, 0.91),     GS56 = c(0.46, 0.92, 0.62, 0.06, 0.87), GS87 = c(0.79, 0.93,     0.86, 0.08, 0.88)), .Names = c("ID", "chr", "gene", "GS12", "GS32", "GS56", "GS87"), class = "data.frame", row.names = c("1", "2", "3", "4", "5"))
df2 <- structure(list(samples = structure(c(1L, 2L, 4L, 3L, 6L, 5L), .Label = c("GS32", "GS33", "GS55", "GS56", "GS68", "GS87"), class = "factor"), ID2 = structure(c(1L, 6L, 3L, 4L, 5L, 2L), .Label = c("GM1", "GM10", "GM17", "GM18", "GM19", "GM7"), class = "factor")), .Names = c("samples", "ID2" ), class = "data.frame", row.names = c(NA, -6L))

Data:

df1:
            ID        chr   gene    GS12      GS32       GS56      GS87
        1 cg29        16    534     0.15      0.44       0.46      0.79  
        2 cg08         3    376     0.87      0.93       0.92      0.93 
        3 cg09         3    171     0.60      0.92       0.62      0.86 
        4 cg65         1    911     0.10      0.07       0.06      0.08
        5 cg36         8    422     0.72      0.91       0.87      0.88
df2:

samples ID2     
GS32    GM1         
GS33    GM7         
GS56    GM17        
GS55    GM18        
GS87    GM19        
GS68    GM10

I would like to subset all columns from df1 (while keeping all the rows in the final output) that are common in ID column of df2, in a nutshell, I would like to subset columns of one data frame according to the rows of another data frame, is there any function that does this?

Try df1[intersect(names(df1), df2$samples)] If df2$samples is factor use as.character(df2$samples) — akrun
I would have a look at the data.table package and the function foverlaps. Maybe this answer given to me would help you too: stackoverflow.com/questions/35719047/… — Phann

Bas Bas · Accepted Answer · 2016-04-15T12:13:29

The input data:

df1 <- structure(list(ID = structure(c(3L, 1L, 2L, 5L, 4L), .Label = c("cg08", "cg09", "cg29", "cg36", "cg65"), class = "factor"), chr = c(16L, 3L, 3L, 1L, 8L), gene = c(534L, 376L, 171L, 911L, 422L), GS12 = c(0.15, 0.87, 0.6, 0.1, 0.72), GS32 = c(0.44, 0.93, 0.92, 0.07, 0.91),     GS56 = c(0.46, 0.92, 0.62, 0.06, 0.87), GS87 = c(0.79, 0.93,     0.86, 0.08, 0.88)), .Names = c("ID", "chr", "gene", "GS12", "GS32", "GS56", "GS87"), class = "data.frame", row.names = c("1", "2", "3", "4", "5"))
df2 <- structure(list(samples = structure(c(1L, 2L, 4L, 3L, 6L, 5L), .Label = c("GS32", "GS33", "GS55", "GS56", "GS68", "GS87"), class = "factor"), ID2 = structure(c(1L, 6L, 3L, 4L, 5L, 2L), .Label = c("GM1", "GM10", "GM17", "GM18", "GM19", "GM7"), class = "factor")), .Names = c("samples", "ID2" ), class = "data.frame", row.names = c(NA, -6L))

I believe what you are asking for is the following:

df1[colnames(df1) %in% df2$samples]
#  GS32 GS56 GS87
#1 0.44 0.46 0.79
#2 0.93 0.92 0.93
#3 0.92 0.62 0.86
#4 0.07 0.06 0.08
#5 0.91 0.87 0.88

You are checking which column names from df1 occur in the samples of df2.
However I assume you also need the ID, chromosome and gene in your output data frame, this can be done with the following:

df1[c(1:3, colnames(df1) %in% df2$samples)]
#    ID chr gene ID.1 ID.2 ID.3
#1 cg29  16  534 cg29 cg29 cg29
#2 cg08   3  376 cg08 cg08 cg08
#3 cg09   3  171 cg09 cg09 cg09
#4 cg65   1  911 cg65 cg65 cg65
#5 cg36   8  422 cg36 cg36 cg36

If you want to force the column order to be in the same order as before, use match instead of %in%. match requires at least two variables, firstone being the target vector, secondone being the data frame/vector which it needs to be sorted to.

df1[,c(1:3,na.omit(match(df2$samples, colnames(df1))))]
#    ID chr gene GS32 GS56 GS87
#1 cg29  16  534 0.44 0.46 0.79
#2 cg08   3  376 0.93 0.92 0.93
#3 cg09   3  171 0.92 0.62 0.86
#4 cg65   1  911 0.07 0.06 0.08
#5 cg36   8  422 0.91 0.87 0.88

Subset columns of one data frame according to another data frame's rows

1 Answers