Error in axis (1, at=ticks)

[bdvelie]

Hi Stephen,

Thanks for developing such a useful tool. I've been using it for awhile and it has always worked great, but I recently ran into a problem that I can't seem to figure out. The manhattan plot function has worked and still works great on my older data sets, but on the newest one it is only working when I limit the data to one chromosome. When I expand it to multiple chromosomes (even only 2) I am continually getting the following error:

Error in axis(1, at = ticks, labels = labs, ...) : no locations are finite In addition: Warning messages: 1: In min(d[d$CHR == i, ]$pos) : no non-missing arguments to min; returning Inf 2: In max(d[d$CHR == i, ]$pos) : no non-missing arguments to max; returning -Inf 3: In min(d[d$CHR == i, ]$pos) : no non-missing arguments to min; returning Inf 4: In max(d[d$CHR == i, ]$pos) : no non-missing arguments to max; returning -Inf Error in axis(1, at = ticks, labels = labs, ...) : no locations are finite

I've tried to sort through it and solve the problem, but I've had no luck. As far as I'm aware the data is in a suitable format and matches all the other files I've used before. At this point I've been staring at it so much even if it were something simple I don't think I'd see it. Any thoughts on what may be causing the problem?

Brandon

[bdvelie]

Forgot to paste a link to some sample data

https://gist.githubusercontent.com/bdvelie/71d9245df7e9a21916f7/raw/Sample

[stephenturner]

Hmm, I'll keep this issue open for now, but here's a temporary fix for you. It looks like the trouble is happening because you only have SNPs for chromosome 3 and 8. I'm not sure off hand if the error is because you're not starting at 1, or if it's because you're "missing" SNPs for 4, 5, 6 and 7. Either way, I should generalize the code more.

But in the meantime, here's a hack. Clobber the data by replacing the 3's with 1s, and the 8s with 2s, draw the plot, but pass a character vector to the chrlabs argument. Like this:

library(qqman)
d <- read.table("Sample.txt", header=TRUE)
d$CHR[d$CHR==3] <- 1
d$CHR[d$CHR==8] <- 2
manhattan(d, chrlabs=c("3", "8"))

[bdvelie]

Hi Stephen,

Thanks for the help. I'm still not sure what was causing the problem yesterday, but the hack works great. Additionally, when I ran it today on the entire dataset (1-31 chr) it worked perfectly fine using the exact same file that was giving the error yesterday. Perhaps it was just something with R that resolved itself.

Thanks again for the help,

Brandon

[zx8754]

I am getting the same error, when I exclude chr1.

[stephenturner]

@zx8754 please post some example data and code as a github gist.

[zx8754]

@stephenturner Here is an example code to reproduce the error:

library(dplyr)
library(qqman)

#using @bdvelie's data
x <- read.table("https://gist.githubusercontent.com/bdvelie/71d9245df7e9a21916f7/raw/Sample",
                header=TRUE) %>% 
  select(CHR,SNP,BP,P)

head(x)
#   CHR          SNP    BP        P
# 1   3 AX-104433016  3525 0.870700
# 2   3 AX-104944723  4679 0.093030
# 3   3 AX-104996406  6794 0.706400
# 4   3 AX-103330248  7003 0.090690
# 5   3 AX-103033987  8340 0.561400
# 6   3 AX-103305412 31255 0.008776

table(x$CHR)
#     3     8 
# 21185 17331 

#add couple of dummy SNPs to chr 1
x_fix <- rbind(
  data.frame(CHR=c(1,1,1),
             SNP=paste0("snp",1:3),
             BP=1:3,
             P=c(0.5,0.003,0.00002)),
  x)

table(x_fix$CHR)
# 1     3     8 
# 3 21185 17331 

manhattan(x) #gives error
manhattan(x_fix) #no problems :)

[stephenturner]

It's a hack, but this works. Obviously need to fix this so manhattan(x) just works.

library(dplyr)
library(qqman)

x <- read.table("https://gist.githubusercontent.com/bdvelie/71d9245df7e9a21916f7/raw/Sample",
                header=TRUE) %>% 
  tbl_df %>% 
  select(CHR,SNP,BP,P)

x %>% arrange(CHR, BP) %>% 
  mutate(realCHRs=CHR) %>% 
  mutate(CHR=factor(CHR, labels=1:n_distinct(CHR)) %>% as.numeric) %>% 
  manhattan(chrlabs=unique(.$realCHRs) %>% as.character)