Issues running MapGen for eQTL finemapping

[llpesce]

I am trying to finemap some eQTL signals, so I am running 3000 or so phenotypes.

It is not always too happy about the LD matrix (my sample is like a 100), but it works OK.

But for one of them I get this error:

selected.sumstats <- gwas.sumstats[gwas.sumstats$locus %in% sig.loci, ]
susie.results <- run_finemapping(selected.sumstats,
region_info = region_info, n=119,
priortype = 'uniform', L= my_L )
sumstats.locus <- gwas.sumstats[gwas.sumstats$locus == sig_loc, ]
susie.sumstats <- merge_susie_sumstats(susie.results, sumstats.locus)
Error in [<-:
! Row index NA is used more than once for assignment.
Run rlang::last_trace() to see where the error occurred.
rlang::last_trace(drop = FALSE)
<error/tibble_error_assign_duplicate_row_subscript>
Error in [<-:
! Row index NA is used more than once for assignment.

---

Backtrace:
▆

1. └─mapgen::merge_susie_sumstats(susie.results, sumstats.locus)
2. ├─base::[<-(*tmp*, snp.idx, "susie_pip", value = <dbl>)
3. └─tibble:::[<-.tbl_df(*tmp*, snp.idx, "susie_pip", value = <dbl>)

4. └─tibble:::tbl_subassign(x, i, j, value, i_arg, j_arg, substitute(value))
   

5.   └─tibble:::vectbl_as_new_row_index(i, x, i_arg, call = call)
   

6.     └─tibble:::abort_assign_duplicate_row_subscript(i, call)
   

7.       └─tibble:::tibble_abort(...)
   

8.         └─rlang::abort(x, class, ..., call = call, parent = parent, use_cli_format = TRUE)
   

I tried to match the sumstats.locus and susie.results manually and I can't find (I extracted like susie.results$25$snp and matched it)

Any ideas? thank you.

[kevinlkx]

Sorry for the delayed response. Thanks for your question. From your code and error, I think the error is likely due to the mismatch between susie.results and sumstats.locus. susie.results contains results for multiple loci (sig.loci), but you used sumstats.locus <- gwas.sumstats$locus == sig_loc, which could be a problem when you have multiple loci in sig.loci.

When you run merge_susie_sumstats(), you should use the susie.results and the same summary stats object used as input for run_finemapping() function. So can you try susie.sumstats <- merge_susie_sumstats(susie.results, selected.sumstats)?