simplest vs bfile_path SAMPLESHEET

[olatzu]

Hello,

Thanks again for the amazing tool and the amazing documentation.

I have been preparing the samplesheet and I believe I am getting the "sampleset" text string wrong.

I have my population in plink.bed plink.fam plink.bim

I understand the "bfile_path" is :

/Users/myname/Desktop/PLINK/plink_mac/plink_genome_test1

Then the "sampleset" could then be "plink_genome_test1"? or how does the sampleset should be looking like?

Thanks a lot for your time in advance and apologies for the inconvenience

Best

[smlmbrt]

An example of the sample sheet is here: https://github.com/PGScatalog/pgsc_calc/blob/main/assets/examples/samplesheet.csv

If your .bed/.fam/.bim files all start with plink_genome_test1 and contain all chromosomes your sample sheet would be:

sampleset,vcf_path,bfile_path,pfile_path,chrom
plink_genome_test1,,/Users/myname/Desktop/PLINK/plink_mac/plink_genome_test1,,

If it is split across chromosomes the files should have slightly longer root names but the same sampleset ID. An example for chrs 1 and 2:

sampleset,vcf_path,bfile_path,pfile_path,chrom
plink_genome_test1,,/Users/myname/Desktop/PLINK/plink_mac/plink_genome_test1_chr1,,1
plink_genome_test1,,/Users/myname/Desktop/PLINK/plink_mac/plink_genome_test1_chr2,,2
[...]

[olatzu]

Dear @smlmbrt ,

Thanks a lot for the quick response! Now worked! The problem seems now to be that "the Score PGS001927_hmPOS_GRCh37 fails minimum matching threshold (22.01% variants match)" is there any way to overcome that? (already tried changing the Ch38 which is even less..) thanks a lot again!

[smlmbrt]

You can adjust the --min_overlap flag to score on the available 22% of variants un the score; however, it's probably best to investigate why the data is missing so many variants. Some options:

• You're only using directly measure genotypes from an array and so the variant overlap is low. This can be solved by running imputation to a common reference panel (e.g. https://imputationserver.sph.umich.edu/) and then supplying the output to the calculator.
• You're using a VCF that doesn't have many samples or proper reference allele encoding. You can see a discussion of that scenario here: format for input data includes reference calls? #50

[olatzu]

Dear @smlmbrt ,

Thank you again very much for the info and the prompt reply! It now seems to have worked. However, I get the error:

Error: --score variant ID '1:89479074:C:T' appears multiple times in main
dataset.

Weird enough I have checked the .bim files and the PGS and the variant is not there(?)

Thank you very much again!

[smlmbrt]

This is because the pipeline relabels the variants for consistency and scoring file formatting. If you do a grep/lookup by position in the .bim file you'll likely see multiple rows with those alleles. Implementing the response in this issue may fix this problem: #74 (comment)