Beagle Gives Error no matter what i do

Question

[Emilie]

I am trying to run genome imputation, so i have my input file and then my phased reference.

I run the following code:

java -jar ${EBROOTBEAGLE}/beagle.jar gt=chr4.vcf.gz ref=Chr4_phased_snp.vcf.gz out=chr4_imputed 

And get the following error

ERROR: REF field is not a sequence of A, C, T, G, or N characters at 4:116707718 [D]

So I trouble shoot, I filter my vcf file so it is only SNPs (so no indels), also try to filter out any sites that do not have A,C,T,G, or N with the following code

bcftools view -i 'REF ~ "^[ACGTN]$"' input.vcf -o filtered_output.vcf

Run it again and get the same thing, so i say hmmm let me look at this specific site within my reference. Well, it does not exist. So just to be safe I run this to get rid of singletons

bcftools view -m2 -M2 -v snps -o output_no_missing.vcf -O v input.vcf.gz

Run beagle again and get the same error, so i say okay lets just use exclude sites with 4:116707718 because it is the problem child so i run

java -jar ${EBROOTBEAGLE}/beagle.jar gt=chr4.vcf.gz ref=Chr4_phased_snp.vcf.gz out=chr4_imputed excludemarkers=file.txt

Where the file has the marker site. I still get the same error. I am out of trouble shooting ideas... It took me forever to find a suitable reference, so I really can't afford to just go find a new one; plus, I am doing this by chromosome, and chromosomes 1-3 worked just fine.

Please help... Please :(

[Pierre Lindenbaum]

you have some answers to validate: Indexing error using Salmon

[Pierre Lindenbaum]

Run it again and get the same thing, so i say hmmm let me look at this specific site within my reference. Well, it does not exist.

what is the output of

bcftools view --targets-overlap 2  --no-header -G --targets "4:116707716-116707720" Chr4_phased_snp.vcf.gz  | cut -f1-5

[Emilie]

That returns nothing as well.

[Pierre Lindenbaum]

4:116707718 [D]

I suspect your VCF is pretty old, in the old days when a deletion was marked 'D'.

can you show me:

bcftools view-G --no-header Chr4_phased_snp.vcf.gz  | cut -f1-5 | awk -F '\t' '!($4~ /^[ATGC]*$/ && $5~ /^[ATGC]*$/)' |head

[Emilie]

The header has the ##filedate=20230306 so I figured it was relatively recent.

That code also returns nothing. It is like beagle is hitting something that doesn't exist, which I know isn't possible but sure feels like it at the moment.

[Pierre Lindenbaum]

duplicate of Error while running BEAGLE for genotype imputation and Imputing missing genotypes using Beagle

[Emilie]

Yes, I saw these and used them for my original troubleshooting. My issue is when I search the vcf file for the location that is in the error, it does not exist in the file. I will keep working on it. Thank you.

[Emilie]

Update, separated out the file for all sites before the 116707718. Confirmed that there was no 116707718 in the file. Ran beagle, got the same error. Its like there is a ghost. Please also note that I have moved on to other chromosomes and have ran beagle just fine with those files