Sorting Chrs In [R]

I should start by saying I have solved this problem, but I feel like my code is ugly and overkill.

I am trying to sort a dataframe on a column containing human chromosomes:

chr1,chr2...chrY, chrX.

the problem is:

chr1, is followed by chr10.

What tricks do you use to deal with this problem IN the [R] environment?

r parsing sort

• 12,811 views • link

updated 14.5 years ago by Jeremy Leipzig

• written 14.5 years ago by Zev.Kronenberg

3 answers

the mixedsort is pretty good, but it will think ChrM comes before ChrX, then ChrY

if you have your own arbitrary order you should just use factors

> df<-data.frame("chr"=c("chr1","chrM","chr10","chr2","chrX","chr2"),"val"=c(1,2,3,4,5,6))
> df
    chr val
1  chr1   1
2  chrM   2
3 chr10   3
4  chr2   4
5  chrX   5
6  chr2   6
> chrOrder<-c(paste("chr",1:22,sep=""),"chrX","chrY","chrM")
> df$chr<-factor(df$chr, levels=chrOrder)
> df$chr
[1] chr1  chrM  chr10 chr2  chrX  chr2 
Levels: chr1 chr2 chr3 chr4 chr5 chr6 chr7 chr8 chr9 chr10 chr11 chr12 chr13 chr14 chr15 chr16 chr17 chr18 chr19 chr20 chr21 chr22 chrX chrY chrM
> df[order(df$chr),]
    chr val
1  chr1   1
4  chr2   4
6  chr2   6
3 chr10   3
5  chrX   5
2  chrM   2

• 1 views • link

written 14.5 years ago by Jeremy Leipzig

thank you for pointing this out.

• 0 views • link

written 14.5 years ago by Gjain

+1 for being in base R, and in the spirit of the R language. Also, this solves @zev.kronenberg issue of sorting: df[order(df$chr, df$pos), ] works as expected. because the factor is sorted based on the underlying integer values. The levels argument to factor() is where the magic happens.

• 0 views • link

written 14.3 years ago by bdemarest

Hi Zev,

what you are looking for is: mixedsort {gtools}

Order or Sort strings with embedded numbers so that the numbers are in the correct order

package 'gtools' was built under R version 2.13.2

n<- c('chr1','chr21','chr13','chr4','chr10')
> n
[1] "chr1"  "chr21"   "chr13"   "chr4"   "chr10"

> mixedsort(n)
[1] "chr1"     "chr4"   "chr10"     "chr13"   "chr21"

I hope this helps.

• 1 views • link

written 14.5 years ago by Gjain

Thanks! data[mixedorder(dat$chromosomes),]

Too bad it wont two vectors. chromosome + position.

• 0 views • link

written 14.5 years ago by Zev.Kronenberg

can you try this way: x<-c('chr2:1-4','chr1:10-15','chr10:2-5','chr5:4-8','chr21:6-23')

mixedsort(x) [1] "chr1:10-15" "chr2:1-4" "chr5:4-8" "chr10:2-5" "chr21:6-23"

• 0 views • link

written 14.5 years ago by Gjain

You could substitute chr1 for chr01 (ie, zero-one), globally, but in that column.

• 1 views • link

written 14.5 years ago by Larry_Parnell

That's not going to help with X, Y or M.

• 0 views • link

written 14.5 years ago by Neilfws

I use this solution frequently, and it does account for X and Y (but not M). Alphabetical sort yields: chr01, chr02, chr10, chr22, chrX, chrY.

• 0 views • link

written 14.3 years ago by bdemarest