Issues with binary dependant variables about collapsabel2

[kindlychung]

Would you mind posting a small dataset so that I can reproduce the error? Thanks!

[mmoisse]

I used your dataset (https://bitbucket.org/kindlychung/collapsabel/downloads/testfiles.zip) to cause the error. If you follow all the terminal and R steps you should get the same errors:

[kindlychung]

Following PLINK conventions, all phenotypes must be numerically coded, including missing values, in the alternate phenotype file. The default missing value is -9.

[kindlychung]

In the case of binary phenotypes, 1 means unaffected, 2 means affected, both 0 and -9 stand for missing. I will update the doc.

[kindlychung]

Please try the latest master and let me if there is a problem.

[kindlychung]

Here is what I tested:

library(CollapsABEL)
rbed_info = rbedInfo(bedstem = "test", TRUE)

pheno <- read.table("test.phe", header=T)
pheno$gc <- pheno$gc + 1
pheno$gc[1] <- -9
write.table(pheno, file="test.12NA.phe", quote=F, row.names=F, col.names=T, sep=" ")

pl_gwas = plGwas(rbed_info,
        pheno = "test.12NA.phe",
        pheno_name = "gc",
        covar_name = "sex,age",
        gwas_tag = "test_gc_sex_age")

gcdh_res = runGcdh(pl_gwas, n_shift = 5)

[mmoisse]

With the latest master it runs fine. Tx!