Troubleshooting

I get an error about variant matching

• Are your target genomes and the scoring file in compatible builds?

• --min_overlap defaults to 0.75 (75% of variants in scoring file must be present in target genomes). Try changing this parameter!

The workflow isn’t using many resources (e.g. RAM / CPU)

Did you forget to set --max_cpu or --max_memory?

You can also edit nextflow.config to configure cpu and memory permanently. nf-core provides a set of example .config files, including examples for both institutional compute clusters (e.g. Cambridge, Sanger) and cloud compute providers (e.g. Google, AWS Tower and Batch). See How do I run pgsc_calc on larger datasets and more powerful computers? for more information.

When I run the workflow I get an error about software not being installed

pgsc_calc bundles dependencies using containers or conda. Did you remember to specify -profile? e.g. nextflow run pgscatalog/pgsc_calc -profile docker,test

Multiple profiles can be combined with a comma. The test profile is used only for checking the pipeline is installed and working correctly.

Cannot access directory error when relabelling genotypes

You might get an error just before one of the following processes finishes:

• PLINK2_RELABELPVAR

• PLINK2_RELABELBIM

• PLINK2_VCF

Error executing process > 'PGSCATALOG_PGSCALC:PGSCALC:MAKE_COMPATIBLE:PLINK2_RELABELPVAR (all_phase3 chromosome ALL)'

Caused by:
  Cannot access directory: <path/to/work/dir>/genomes/all_phase3/ALL

This is caused by certain local storage configurations.

Rerunning the pipeline with -resume will fix the problem.

To avoid the problem happening, set the --genotypes_cache parameter to a directory that already exists on your file system.

I’m having problems with VCF input

If you use a “chr” prefix in the chromosome column of your VCF, please remove it. Here’s a simple method to do this (thanks to Rvtests):

(zgrep ^"#" $your_old_vcf; zgrep -v ^"#" $your_old_vcf | sed 's:^chr::ig' | sort -k1,1n -k2,2n) | bgzip -c > $your_vcf_file.gz

VCF file(s) containing variants on non-standard chromsomes or patches (e.g. chr1_gl000191_random) will also currently fail our pipeline as it only takes human chromosomes as input (1-22, X, Y, XY). One way to remove these variants is to download and run plink2 and convert your data to plink files that can be used with the calculator using the following command:

plink2 --vcf [yourfile] --allow-extra-chr --chr 1-22, X, Y, XY -make-pgen --out [yourfile]_axy

however other methods to filter these variants from VCFs also exist.

By default the pipeline uses the genotypes present in the GT field of the VCF file. If you would like to use imputed dosages you must add a vcf_genotype_field field column to the samplesheet with the DS value. See How to set up a samplesheet for more information.