"NA or NaN values in the resulting correlation matrix."

[JNajar]

Hi!

I have used your script for LDPred2 several times with good results. However, now that i want to create a PRS for diabetes type 2, for some reason i get this warning during the LD calculations: "Warning message: NA or NaN values in the resulting correlation matrix." I have already removed NAs in a previous step:

# a lot of NAs for hg38 variable 
sumstats <- na.omit(sumstats)

And when i check for NAs i dont find any. 

If i continue with the script i get this: 
> ## We assume the fam order is the same across different chromosomes
> fam.order <- as.data.table(obj.bigSNP$fam)
>
> ## Rename fam order
> setnames(fam.order,
+         c("family.ID", "sample.ID"),
+         c("FID", "IID"))
>
> # 6. Perform LD score regression
>   df_beta <- info_snp[,c("beta", "beta_se", "n_eff", "_NUM_ID_")]
>
>   ldsc <- snp_ldsc(   ld,
+                       length(ld),
+                       chi2 = (df_beta$beta / df_beta$beta_se)^2,
+                       sample_size = df_beta$n_eff,
+                       blocks = NULL)
Error in if (max(abs(pred - pred0)) < 1e-06) break :
  missing value where TRUE/FALSE needed
>
>   h2_est <- ldsc[["h2"]]
Error: object 'ldsc' not found

Do you know what might be the problem? Or how I can troubleshoot?

All the best, Jenna

[privefl]

Have you searched for similar issues? I think that has been reported a few times here already.

[privefl]

Please make sure to post your issues in the bigsnpr repo.

[JNajar]

Pretty new to Github, sorry :) I am not sure how?

[privefl]

[Minta821]

I have also used your script for LDPred2 several times, I also got the same error now .

[privefl]

Yes, the change appeared in v1.9.5 (cf. https://privefl.github.io/bigsnpr/news/):

snp_cor() and bed_cor() now return NaNs when e.g. the standard deviation is 0 (and warn about it). Before, these values were not reported (i.e. treated as 0).

So, yes, it "worked" before, but for wrong reasons. Please make sure to do some proper QC on very small MAF and this warning will disappear.

[privefl]

Any update on this?

[Minta821]

Any update on this?

Sorry for the late responses, yes, it's working after SNP filtering..