PLINK - PRS/GRS .fam file has fewer tokens

Question

[Jalil Sharif]

Hi Everyone,

I am this error for developing a PRS.

Error: Line 2 of .fam file has fewer tokens than expected.

My script looks as follows:

/rds/general/user/js4120/home/bin/plink \
    --bfile merged_chr.QC.munged \
    --fam master_combined.fam \
    --keep fid_iid_keep \
    --score GWAS_final.qc 3 4 6 header double-dosage no-mean-imputation \
    --q-score-range range_list SNP.munged.pvalue \
    --extract valid.snp.munged \
    --out chr.QC.munged.grs.pvalue

The .fam file for all previous steps (filtering and clumping) looked like this:

-1  -1  0   0   0   -9
-2  -2  0   0   0   -9
-3  -3  0   0   0   -9
1000016 1000016 0   0   0   -9
1000028 1000028 0   0   0   -9
1000033 1000033 0   0   0   -9
1000042 1000042 0   0   0   -9
1000051 1000051 0   0   0   -9
1000069 1000069 0   0   0   -9
1000075 1000075 0   0   0   -9
...

total lines: 487409

As I did not have any gender or phenotype information, I changed the .fam file, to include gender and phenotype, so my new .fam file looks like this.

#FID    IID SEX Phenotype_combined
-2  -2  -9  -9
-3  -3  -9  -9
1000016 1000016 2   1
1000028 1000028 2   1
1000033 1000033 2   1
1000042 1000042 2   1
1000051 1000051 2   1
1000069 1000069 2   1
1000075 1000075 2   1
...
total lines 487409

I wanted to clarify two things,

1st: Does my new .fam file have to have six columns? e.g.

#FID    IID SEX Phenotype_combined NA NA
-2  -2  -9  -9  -9  -9
-3  -3  -9  -9  -9  -9
1000016 1000016 2   1   -9  -9
1000028 1000028 2   1   -9  -9

Secondly changing the .fam file between clumping and scoring, impact on the result of the PRS?

E.g. my clumping step didn't have case/control information, but now with the PRS it will.

Many thanks