Model training. Convert VCF file to PrediXcan genotype dosage format

[Anh Do]

Hi all,

Do you know if any PrediXcan script available to convert VCF file into PrediXcan genotype dosage to run model training?

The PrediXcan genotype dosage described at 

https://github.com/hakyimlab/PredictDB-Tutorial/wiki/File-descriptions

with variants in rows and samples in columns.

Do we suppose to use filter_and_covert()? It doesn't work for me.  I got the error message as below.

"[W::bcf_sr_add_reader] No BGZF EOF marker; file '*.vcf.gz' may be truncated
[W::bcf_hdr_check_sanity] GL should be declared as Number=G
[W::vcf_parse] Contig '1' is not defined in the header. (Quick workaround: index the file with tabix.)
[W::vcf_parse_format] FORMAT 'GT' is not defined in the header, assuming Type=String
Undefined tags in the header, cannot proceed in the sample subset mode.
[W::bcf_hdr_check_sanity] GL should be declared as Number=G"

Thanks very much.

Anh

[Festus]

Hi Anh,

You can use this approach.

VCF > Plink BED > Plink Text format.

Using the plink --recode A-transpose generates an additive file which you can can load into R together with the plink bim file to format the data into the format required to train the model.

Regards,

Festus

[Festus]

The filter and convert is a custom function here which you can use too.

[Anh Do]

Thanks, Festus. The Plink approach works nicely.

Anh