This question already has an answer here:
-
How can I prune these column ranges given another file in R?
1 answer
I have a df1 called ranges like:
1 bin chrom chromStart chromEnd name score
2 12 chr1 836780 856723 -5.7648 599
3 116 chr1 1693001 1739032 -4.8403 473
4 117 chr1 1750780 1880930 -5.3036 536
5 121 chr1 2020123 2108890 -4.4165 415
I also have a data.frame called viable like:
chrom chromStart chromEnd N
chr1 840000 890000 1566
chr1 1690000 1740000 1566
chr1 1700000 1750000 1566
chr1 1710000 1760000 1566
chr1 1720000 1770000 1566
chr1 1730000 1780000 1566
chr1 1740000 1790000 1566
chr1 1750000 1800000 1566
chr1 1760000 1810000 1566
Basically I have ranges of values from chromStart to chromEnd in ranges. I also have a list of ranges in the df2 viable. The ranges in viable are much smaller. I want to test the ranges from ranges and make sure that the full range falls within ranges that are viable. How can I do this?
The output I want is a data.frame like:
1 bin chrom chromStart chromEnd name score
2 12 chr1 840000 856723 -5.7648 599
3 116 chr1 1693001 1739032 -4.8403 473
6 133 chr1 1750780 1880930 -4.8096 469
You could try using the GenomicRanges package.
library(dplyr)
library(GenomicRanges)
Here we load in the the example input data. (This is an inelegant way to do this -- I know... but I was lazy and the sublime multiline edit made it easy.) Note: I don't know where the "1" column means, but I kept it in the data.
ranges <-
rbind(
c("2","12","chr1","836780","856723","-5.7648","599"),
c("3","116","chr1","1693001","1739032","-4.8403","473"),
c("4","117","chr1","1750780","1880930","-5.3036","536"),
c("5","121","chr1","2020123","2108890","-4.4165","415")
) %>%
as.data.frame()
colnames(ranges) <-
c("1","bin","chrom","chromStart","chromEnd","name","score")
viable <-
rbind(
c("chr1","840000","890000","1566"),
c("chr1","1690000","1740000","1566"),
c("chr1","1700000","1750000","1566"),
c("chr1","1710000","1760000","1566"),
c("chr1","1720000","1770000","1566"),
c("chr1","1730000","1780000","1566"),
c("chr1","1740000","1790000","1566"),
c("chr1","1750000","1800000","1566"),
c("chr1","1760000","1810000","1566")
) %>%
as.data.frame()
colnames(viable) <-
c("chrom","chromStart","chromEnd","N")
## Need columns to be integers
ranges <-
ranges %>%
tbl_df() %>%
mutate(
chromStart = chromStart %>% as.character %>% as.integer,
chromEnd = chromEnd %>% as.character %>% as.integer
)
viable <-
viable %>%
tbl_df() %>%
mutate(
chromStart = chromStart %>% as.character %>% as.integer,
chromEnd = chromEnd %>% as.character %>% as.integer
)
Here is where my answer begins:
- Reformat dataframe to GenomicRanges class
- Find the regions by doing an intersection
- Add in the bin, name, and score columns using the
findOverlaps. (Note, this information is removed during the intersection because there is not necessarily a 1:1 mapping)
- Reformat output back into a dataframe
Done
gr.ranges <-
makeGRangesFromDataFrame(ranges,
keep.extra.columns = T,
seqnames.field = "chrom",
start.field = "chromStart",
end.field = "chromEnd")
gr.viable <-
makeGRangesFromDataFrame(viable,
keep.extra.columns = T,
seqnames.field = "chrom",
start.field = "chromStart",
end.field = "chromEnd")
# To find the intersects
gr.intersect <-
GenomicRanges::intersect(gr.ranges, gr.viable)
# For linking up the non- chrom,start,end columns
gr.hits <-
GenomicRanges::findOverlaps(gr.intersect, gr.ranges)
output <-
gr.intersect[queryHits(gr.hits)]
mcols(output) <-
mcols(gr.ranges[subjectHits(gr.hits)])
output
# Reformat to dataframe
output %>%
as.data.frame() %>%
select(`1` = X1, bin, chrom = seqnames, chromStart = start, chromEnd = end, name, score)
