CODEWITHSUNDEEP
rdataframe
Sander Van der Zeeuw
Sander Van der Zeeuw

Reputation: 1092

Rank columns and put the column with the highest score as first and so on

Lets say i have a very large data.frame containing scores per column.

for example:

MA0001.1 AGL3 MA0003.1 TFAP2A MA0004.1 Arnt  MA0005.1 AG   MA0006.1 Arnt::Ahr
7.789524e-09  0.4012127249    3.771518e-03   1.892011e-06  0.002733200
5.032498e-07  0.0001873801    9.947449e-05   3.284222e-05  0.001367041
1.194487e-06  0.0009357406    6.943634e-05   1.589373e-05  0.002551519
4.833494e-06  0.0150703600    1.003488e-04   1.197928e-03  0.001431416
6.865040e-05  0.0000732607    3.857193e-04   5.388744e-03  0.001363706

R data.frame:

testfr<-structure(list(`MA0001.1 AGL3` = c(7.78952366977488e-09, 5.03249791215203e-07, 
1.19448739380034e-06, 4.83349413748598e-06, 6.86504034402563e-05
), `MA0003.1 TFAP2A` = c(0.401212724871542, 0.000187380067026448, 
0.000935740631438077, 0.0150703600158589, 7.32607018758816e-05
), `MA0004.1 Arnt` = c(0.00377151826447817, 9.94744903768433e-05, 
6.94363387424972e-05, 0.000100348764966112, 0.00038571926458373
), `MA0005.1 AG` = c(1.89201084302835e-06, 3.2842217133538e-05, 
1.58937284554136e-05, 0.00119792816070882, 0.00538874414923338
), `MA0006.1 Arnt::Ahr` = c(0.00273319966783363, 0.00136704060025893, 
0.00255151921946167, 0.00143141576426544, 0.00136370552325235
)), .Names = c("MA0001.1 AGL3", "MA0003.1 TFAP2A", "MA0004.1 Arnt", 
"MA0005.1 AG", "MA0006.1 Arnt::Ahr"), class = "data.frame", row.names = c(4L, 
2L, 5L, 1L, 3L))

Now i want to select the column with the highest values in it and place that column first. So the values of 1 column should stay below the same column name and the entire column should move by rank.

I tried the following:

ranked<-unlist(lapply(testfr,rank))
testranked<-testfr[ranked, ]

this produces a data frame with 2259obs*459vars while the original was 5*459.

Note that, testfr is a data.frame derived from a function which scores sequences on to a list of matrices! And gives that score back into a data.frame where the rows are the sequences and the columns are the matrices.

I know i do something wrong with the indexing or unlisting but i dont have any clue how to fix this. Any help is appreciated.

Upvotes: 3

Views: 620

Answers (2)

agstudy
agstudy

Reputation: 121568

I would use apply for readability, 

testfr[order(apply(testfr, 2, max, na.rm = TRUE),decreasing=T)]

I apply max for each margin , column here, Then I sort column in decreasing order.

Upvotes: 1

Arun
Arun

Reputation: 118799

How about this?

> testfr[rev(order(sapply(testfr, max, na.rm = TRUE)))]

Break down:

sapply(test.fr, max, na.rm = TRUE) # get max of each column (after removing NA)
order(.) # get the order of these values in increasing order
rev(.)   # get the reverse order so that highest value index stays first
testfr[.] # get the columns in this order back

Upvotes: 7

Related Questions