a99_plink_cheatsheet.Rmd| PLINK | tidypopgen |
|---|---|
| –make-bed –out | gt_as_plink(data, bedfile = my_file) |
| –recode | `gt_as_plink(data, bedfile = my_file |
| –allele1234 and –alleleACGT | See gen_tibble()\ parameter ‘valid_alleles’ |
PLINK flags –update-alleles, –allele1234, and –alleleACGT, all alter
the coding of alleles. In tidypopgen, valid alleles are
supplied when reading in a gen_tibble.
| PLINK | tidypopgen |
|---|---|
| –maf | data %\>% loci_maf() |
| –geno | data %\>% loci_missingness() |
| –mind | data %\>% indiv_missingness() |
| –hwe | data %\>% loci_hwe() |
| –het | data %\>% indiv_het_obs() |
| –distance | data %\>% gt_ibs() |
To filter out variants in tidypopgen, in a similar way
to PLINK flags such as –extract or –autosome, it is necessary to use the
gen_tibble with select_loci_if(). For
example:
data %>% select_loci_if(loci_chromosomes(genotypes) %in% c(1:22))
#> # A gen_tibble: 15 loci
#> # A tibble: 5 × 3
#> id population genotypes
#> <chr> <chr> <vctr_SNP>
#> 1 GRC24 pop_a 1
#> 2 GRC25 pop_a 2
#> 3 GRC26 pop_a 3
#> 4 GRC27 pop_a 4
#> 5 GRC28 pop_a 5will select autosomal loci in the same way as –autosome. Or alternatively:
my_snps <- c("rs4477212","rs3094315","rs3131972","rs12124819","rs11240777")
data %>% select_loci_if(loci_names(genotypes) %in% my_snps) %>% show_loci()
#> # A tibble: 5 × 7
#> big_index name chromosome position genetic_dist allele_ref allele_alt
#> <int> <chr> <int> <int> <int> <chr> <chr>
#> 1 1 rs4477212 1 82154 0 A NA
#> 2 2 rs3094315 1 752566 0 A G
#> 3 3 rs3131972 1 752721 0 G A
#> 4 4 rs12124819 1 776546 0 A NA
#> 5 5 rs11240777 1 798959 0 G Awill select loci from a previously defined set in the same way as –extract.
Similarly, to filter out individuals, as might be performed with –keep in PLINK, requires using filter:
Linkage is managed through clumping in tidypopgen with
loci_ld_clump().
This option is similar to the –indep-pairwise flag in PLINK, but results in a more even distribution of loci when compared to LD pruning.
To explore why clumping is preferable to pruning, see https://privefl.github.io/bigsnpr/articles/pruning-vs-clumping.html
| KING | tidypopgen |
|---|---|
| –kinship | gt_king() |
| –unrelated | filter_high_relatedness() |
gt_king() implements the KING-robust estimator of kinship equivalent to –kinship in KING. To remove related individuals, instead of using –unrelated –degree, the user can pass the resulting kinship matrix and a relatedness threshold to filter_high_relatedness, which will return the largest possible set of unrelated individuals.
| PLINK | tidypopgen |
|---|---|
| –bmerge | rbind() |
| –flip-scan | rbind_dry_run() |
| –flip | use ‘flip_strand = TRUE’ in rbind()
|
In PLINK, data merging can fail due to strand inconsistencies that
are not addressed prior to merging. PLINK documentation suggests to
users to try a ‘trial flip’ of data to address this, and then to
‘unflip’ any errors that remain. In tidypopgen, when data
are merged with rbind, strand inconsistencies are identified and
automatically flipped, avoiding multiple rounds of flipping before
merging.
PLINK does allow users to identify inconsistencies prior to merging
with –flip-scan, and this functionality is included in the
tidypopgen rbind_dry_run(). rbind_dry_run() reports the
numeric overlap of datasets, alongside the number of SNPs to ‘flip’ in
the new target dataset, as well as the number of ambiguous SNPs.
Data are only merged one set at a time, there is no equivalent to –merge-list.