str score error for some loci

[krukanna]

Hi,
I've got error when running:
tsum <- tsum_test(str_score)
For locus like DBQD2, EIEE1, FRAXE and many other got message:

'apply(qm, 2, sort.int, partial = ti)': dim(X) must have a positive length

or

'if (qs < 1) {': argument is of length zero

To avoid the error I have to remove these locus from the function like below:
tsum <- tsum_test(str_score[c("DM2","DRPLA","FRA12A","FRAXA","GDPAG")]

Do you know what could be the cause?
Thanks for help

[trickytank]

Hi,
This is likely due to low coverage at those loci.
I have fixed this in a development branch, but that version isn't passing all tests at the moment. I'll push that to a beta branch once I've fixed it, and I'll tell you how to install it. Should take about a week.

[trickytank]

Hi @krukanna, you can try install using this command in R:

devtools::install_github("bahlolab/exSTRaDevelopment", "beta")

Let me know if you still have an issue.

[krukanna]

Hi, sorry for the delay.
I used this beta version and this error is fixed. Now I've got information:
Not enough data (low counts)
That's fine, it doesn't break the script.

But I also got new error:

Error in command 'if (verbose && p.value.sd < early_A * p_smallest) {':
  values are missing where TRUE/FALSE is required
And warnings:
1: In command 'tsum_test(str_score)':
  Trimming at each quantile only leaves 3 observations at each level.
2: In command 'sqrt(bac_var * (nrow(qmt_bac) + 1)/(nrow(qmt_bac) - 1))':
  NaN values were produced

[krukanna]

Below is my workflow. I assume that at the end in the variable p_signif I will get a locus or more that can be suspected of repeat expansion. Should I have a minimum number of samples that I process together or it doesn't matter?

str_score <- read_score (
  file = "exSTRa_scores.txt", 
  database = "repeat_expansion_disorders_grch38.txt",
  groups.regex = c(control = "^CP000.$", case = "")
)
sum <- tsum_test(str_score[locus != c("DRPLA","SCA1"),])
p <- p_values(tsum, correction = "samples")
p_signif <- p_values(tsum, only.signif = TRUE, correction = "samples")

Thanks for answering.
Best,
Anna