Skip to contents

Comparison of rsid vs chr:pos lookup in ElasticSearch

Gibran Hemani

Source: vignettes/timings.Rmd
timings.Rmd

Get a set of SNPs to query:

a <- tophits("ieu-a-2")
rsid <- a$rsid
chrpos <- paste0(a$chr, ":", a$position)
length(rsid)

1. Comparison of rsid lookup vs chromosome:position 

mbm <- microbenchmark("rsid" = {
        b <- associations(rsid, "ieu-a-7", proxies=0)
    },
    "chrpos" = {
        b <- associations(chrpos, "ieu-a-7", proxies=0)
    }, times=10
)
kable(summary(mbm))

2. Single range query

Specify a range, and then get the corresponding rsIDs in that range

radius <- 100000
chrpos <- paste0(a$chr[1], ":", a$position[1]-100000, "-", a$position[1]+100000)
b <- associations(chrpos, "ieu-a-7", proxies=0)
rsid <- b$rsid

Window size

format(radius * 2, scientific = FALSE, big.mark = ",")
length(rsid)
mbm <- microbenchmark("rsid" = {
        b <- associations(rsid, "ieu-a-7", proxies=0)
    },
    "chrpos" = {
        b <- associations(chrpos, "ieu-a-7", proxies=0)
    }, times=10
)
kable(summary(mbm))

3. Multiple range queries

Specify multiple ranges, and then get the corresponding rsIDs across all those ranges

chrpos <- paste0(a$chr, ":", a$position-10000, "-", a$position+10000)
b <- associations(chrpos, "ieu-a-7", proxies=0)
rsid <- b$rsid
length(chrpos)

ranges of

format(20000, scientific = FALSE, big.mark = ",")

size window, which is covered by

length(rsid)

variants to lookup.

mbm <- microbenchmark("rsid" = {
        b <- associations(rsid, "ieu-a-7", proxies=0)
    },
    "chrpos" = {
        b <- associations(chrpos, "ieu-a-7", proxies=0)
    }, times=5
)
kable(summary(mbm))