snpR

snpR is an R package for analyzing call Single Nucleotide Polymorphism (SNP) genotypes containing most basic stats including pairwise LD, gaussian sliding window analysis tools, plotting options, clustering analysis, colony interface, Ne estimation, formatting, filtering, and more! It is built primarily to be user-friendly and handle many levels of SNP and sample metadata without the need for complicated file or object management. Please see the example below!

Installation

snpR can be installed from GitHub:

# install.packages("remotes")
remotes::install_github("hemstrow/snpR")

To install the vignettes as well (recommended for new users), instead use:

remotes::install_github("hemstrow/snpR", build_vignettes = T) # linux
remotes::install_github("hemstrow/snpR", build_vignettes = T, build_opts = c("--no-resave-data", "--no-manual")) # windows

If you wish to try out the latest features or bug fixes, the dev version can be installed from GitHub as well:

# install.packages("remotes")
remotes::install_github("hemstrow/snpR", ref = "dev")

A CRAN version should be available soon.

Function Table of Contents

Import:

• import.snpR.data(): generic read function, takes many file types by extension or R data.frames().
• Wrappers for specific file types:
  • read_structure(): Reads STRUCTURE “.str” files.
  • read_vcf(): Reads VCF “.vcf” files.
  • read_FSTAT(): Reads FSTAT “.fstat” files.
  • read_ms(): Reads ms “.ms” files.
  • read_delimited_snps(): Reads tab-delimited “NN” or “0000” data.
  • read_genepop(): Reads genepop “.genepop” files.
  • read_plink(): Reads PLINK! “.bed”, “.fam”, and “.bim” files.
  • convert_genlight(): Converts adegenet genlight class objects.
  • convert_genind(): Converts adegenet genind class objects.
  • convert_vcfR(): Converts vcfR class objects.

Utility:

• filter_snps(): Filter data.
• format_snps(): Format data into other export formats.
• summarize_facets(): Summarized available facets.
• citations(): Fetch citations for all methods used in calculations for a specific snpRdata object.
• check_duplicates(): Check data for potentially duplicated samples.
• gap_snps(): Select a SNP every n bases (simple physical LD filtering).
• vcf2beagle() and vcf2PL(): Utilities to convert vcf files to either beagle or pl/mgpl files. The latter can also optionally prepare inputs for the program entropy.

Object Access and Manipulation:

• Dimensions:
  • nsnps() and nrow(): Get the number of SNPs in an object.
  • nsamps() and ncol(): Get the number of samples in an object.
  • dim(): Get number of SNPs and samples in an object.
• Access:
  • get.snpR.stats(): Fetch any calculated statistics from an object.
  • genotypes(): Fetch genotypes.
  • sample.meta(): Fetch (or reasign with <-) sample metadata.
  • snp.meta(): Fetch (or reasign with <-) SNP metadata.
• Subetting:
  • [: The usual bracket operator. Subset by SNP or sample index, or by facet.
  • subset_snpR_data(): Wrapper for the bracket operator.

Statistics:

• Basic statistics
  • calc_pi(): Nucleotide diversity.
  • calc_ho(): Observed heterozygosity.
  • calc_he(): Expected heterosygosity.
  • calc_hwe(): Hardy-Weinburg equilibrium (HWE).
  • calc_hs(): Standardized individual heterozygosity.
    • calc_het_hom_ratios(): Alternative, raw heterozygote/homozygote ratios within individuals.
  • calc_ne(): Effective population size.
  • calc_prop_poly(): The proportion of polymorphic loci.
  • calc_maf(): Minor allele frequencies, calculated automatically when any facet operations are performed.
  • calc_private(): Rarefaction-corrected detection of private alleles across facet levels.
  • calc_seg_sites(): Rarefaction-corrected estimates of the number of segregating sites per facet level.
  • calc_allelic_richness():: Rarefaction-corrected estimates of allele counts per locus per facet level.
  • calc_genetic_distances(): Genetic distances between individuals.
  • calc_fis(): $F_{IS}$ (inbreeding coefficients).
  • calc_pairwise_fst(): Pairwise $F_{ST}$ between facet levels.
  • calc_global_fst(): Global $F_{ST}$ across facet levels.
  • calc_pairwise_ld(): Pairwise LD between SNPs.
  • calc_abba_baba(): ABBA/BABA tests.
  • calc_tajimas_d(): Tajima’s D globally (or across windows). Also calculates Watterson’s and Tajima’s $\theta$.
• Association:
  • calc_association(): Association testing against a phenotype.
  • run_random_forest(): Run a random forest prediction/association test against a phenotype.
  • run_random_forest(): Run genomic prediction against a phenotype.
    • cross_validate_genomic_prediction(): Bare-bones cross-validation for genomic predictions.
• Site-frequency Spectra:
  • calc_sfs(): Generate a 1 or 2d site frequency spectra.
    • make_sfs(): Wrapper function that uses an external dadi formatted file to generate an sfs.
  • calc_directionality(): Peter and Slatkin’s directionality index.
  • calc_origin_of_expansion(): Estimate the origin point of a range expansion based on directionality.
• Other:
  • calc_isolation_by_distance(): Run an IBD mantel test.
  • calc_tree(): Generate a tree based on individual or facet-level relatedness.
  • tabulate_allele_frequency_matrix(): Generate an allele frequency matrix.

Windows:

• calc_smoothed_averages(): Core function to do sliding window analysis using a gaussian smoothing kernal.
• calc_tajimas_d(): Tajima’s D across sliding windows (or globally). Also calculates Watterson’s and Tajima’s $\theta$.
• Bootstrapping:
  • do_bootstraps(): Core function to generate bootstrapped significance values for smoothed windows (elevation or reduction vs genomic background).
    • calc_p_from_bootstraps(): Calc p-values from bootstraps. Run automatically by do_boostraps().

Plotting:

• plot_clusters(): PCA, UMAP, and tSNE plots.
• plot_structure(): Run STRUCTURE or several alternatives OR read in existing “q” files and generate plots.
• plot_structure_map(): Plots plot_structure() or parsed in q file results on a map given coordinates for populations.
• plot_diagnostic(): A suite of useful diagnostic plots.
• plot_manhattan(): Manhattan plots from calculated statistics or a data.frame(). Excellent for visualizing most statistics genome-wide (not just association tests!)
• plot_qq(): Quantile-quantile (qq) plots from calculated association test results.
• plot_pairwise_fst_heatmap(): Heatmap of FST scores between facet levels.
• plot_pairwise_ld_heatmap(): Heatmap of LD scores between SNPs.

Parentage:

• Colony:
  • run_colony(): All-in-one function to make a colony import file, run colony, and parse results.
  • write_colony(), call_colony(), parse_colony(): Write input files, call colony, and parse results as seperate functions.
• Sequoia:
  • run_sequoia(): Run a basic parentage assessment with the sequoia package.

Example

snpR is focused on ease-of-use. Primarily, it achieves this via the use of facets, which describe sample or SNP metadata. snpR is built to automatically split up analysis by facet. For example, calculating observed heterozygosity for each population or family, or for each population/family combination is easy!

Statistic calculation with facets

library(snpR)
## basic example code

x <- calc_ho(stickSNPs, facets = c("pop")) # split by pop (stickSNPs is an example dataset included in snpR)
x <- calc_ho(x, facets = c("fam")) # split by family
x <- calc_ho(x, facets = c("pop.fam")) # split by combinations of family and pop

snpR also facilitates ease-of-use by being overwrite safe. As above, new analyses are added to an existing object. Results can be fetched using the get.snpR.stats handler.

res <- get.snpR.stats(x, facets = "pop", stats = "ho")

Functions in snpR are consistently named: functions that calculate statistics are prefixed calc_, functions that do plots are prefixed plot_, and functions that run external tools (like COLONY), are named run_. Typing snpR::calc into the console on Rstudio will bring up a helpful list of all of the statistical functions!

Basic plotting

snpR provides a suite of plotting tools, all of which are named plot_...(). For example, a PCA can be plotted using plot_clusters():

plot_clusters(x, "pop", plot_type = "pca")

Citation tools

snpR automatically tracks citations for all of the methods used on a snpRdata object and can provide them or write a .bib file on request using the citations() function:

citations(x,
          outbib = FALSE, # if TRUE, writes a .bib file for all methods
          return_bib = FALSE # if TRUE, returns a list of .bib entries
          )

Vignette

For a full introduction, check the snpR_introduction vignette.

# remotes::install_github("hemstrow/snpR", build_vignettes = T, build_opts = c("--no-resave-data", "--no-manual"))

vignette("snpR_introduction")