tidypopgen 1.0.0

New features and changes

• optimisations to work on whole genomes, both speed and memory
• new functions to compute site frequency spectra (SFS) for demographic inference: pop_sfs(),pairwise_pop_sfs() and nwise_pop_sfs()
• native windows_indiv_roh() matching the algorithm used by PLINK 1.9 (--homozyg); the old version, based on detectRUNS, is deprecated as gt_roh_window()
• gt_roh_froh(), gt_roh_loci() and gt_roh_islands() summarise runs of homozygosity (F_ROH, loci in runs, and ROH islands)
• jackknife_* functions for genome-wide statistics with block jackknife standard errors
• pop_tajimas_d() and related functions handle missing genotypes as pixy, and work with polyploid and pseudohaploid data
• autoplot() method for the distance matrices of pairwise_* functions
• option to use integer loci names (names_as_int) for large datasets, when reading VCF and PLINK bed files
• polyploid support in indiv_het_obs(), pop_het_obs(), loci_pi() and gt_as_genind(); clear errors where polyploids are not supported
• pairwise_ibs(type = "adjusted_counts") now returns PLINK’s allele count distance (--distance flat-missing)
• pairwise_* functions use imputed genotypes when set with gt_set_imputed(), and use several cores by default
• gen_tibble() on ped files stores individual information in the same columns as for bed files (see ?gen_tibble)
• gen_tibble() copies the backing file of bigSNP objects (never modifying them), and works on read-only files
• rbind(use_position = TRUE) keeps the loci names of the reference, and needs much less memory; integer names are matched by position
• the as_is argument of rbind() is deprecated (it was never implemented)
• cbind() renames duplicated columns as dplyr::bind_cols(); id and genotypes can not be added
• distruct_colours holds the 56 distinct distruct colours, recycled with a warning by scale_fill_distruct()
• gt_as_genlight() keeps alleles, chromosomes and positions, so that gt_from_genlight() can convert back
• classes are preserved by all dplyr verbs; grouped functions keep the order of factor levels, and work with several grouping variables and empty groups
• heatmap_add_dendro() returns a gtable, which can be combined with patchwork
• augment_loci() works on gt_pca_autoSVD() results
• gt_admixture() uses ADMIXTURE on the PATH when conda is not installed
• faster “cpp” VCF parser, and select_loci_if() no longer loads all the genotypes
• informative errors for invalid inputs to gen_tibble(), show_loci<-, gt_order_loci(), gt_pcadapt(), gt_dapc(), pop_fst() and others

Bug fixes

• gen_tibble(): fix backing file names (truncated at dots, or “.vcf.bk”), and backing files left behind on errors
• gen_tibble(): fix files with trailing blank lines, CRLF line endings or upper case extensions
• gen_tibble() on bed files keeps .fam IDs as text, and quantitative phenotypes as numbers
• gen_tibble() on VCF files: fix the ploidy of individuals missing on the first variant, partially missing genotypes, and genotypes above the ploidy
• gen_tibble(): “0” and “.” are always missing alleles; fix the “vcfR” parser on indels and on missing genotypes on the second variant
• gt_load(): fix the ploidies of gen_tibbles saved in the old bigSNP format
• gt_as_plink(): fix the .fam and .map information, missing alleles, and ped/raw files with numeric phenotypes, no population or pseudohaploids
• gt_as_vcf() and gt_as_plink(): fix positions in scientific notation, and the date in the VCF header
• gt_admixture() and gt_snmf(): fix seeds with several runs, paths containing “geno”, and the order of loci in P matrices
• gt_snmf(): fix continued projects, and the order of cross-entropies
• gt_impute_xgboost(): fix unsorted loci; gt_impute_simple(): restore the BLAS option
• filter_high_relatedness(): fix missing and strongly negative kinship coefficients
• snp_allele_sharing(): fix blocks of different sizes; indiv_het_obs(): fix the number of loci after select_loci()
• pop_global_stats(): fix loci where a population has no data; pop_tajimas_d(): fix loci with no valid genotypes
• loci_alt_freq() and loci_maf(): fix polyploids and single individuals
• grouped loci_* functions and qc_report_*: fix factor, numeric, non-contiguous and empty groups, and p-values above 1
• loci_transitions() and loci_transversions(): fix misclassified substitutions
• windows_* functions: fix the last loci of chromosomes; windows_pairwise_pop_fst() and windows_nwise_pop_pbs() ignoring method
• gt_roh_window(): fix homozygotes for the alternate allele with heterozygosity = TRUE
• gt_dapc() and gt_cluster_pca(): fix the default number of components, the variance retained, and the choice of k
• gt_pca_autoSVD(): fix unsorted loci; augment() of gt_pca and gt_dapc: fix single components and integer ids
• rbind(): fix ploidies after filtering, and missing allele_ref; rbind_dry_run(): fix paths to .bim files
• dplyr verbs on grouped gen_tibbles: fix select(), rename(), relocate(), [, ungroup(), group_by() and mutate()
• autoplot() of q_matrix and gt_admix: fix the sorting options, numeric and unused groups, and type = "cv"; ordering in other autoplot()s
• gt_admix_reorder_q(): fix K = 1
• heatmap_add_dendro(): fix ties and older ggplot2; heatmap_pairwise(): fix factor ids
• fix chromosome names being renumbered in lexicographic order
• gt_pseudohaploid(): fix other individuals of the backing file, and error on data that are not diploid
• tidy() and augment() of q_matrix: fix ignoring individual ids
• gt_as_genind(): fix loci names with “.”; gt_from_genlight(): fix objects without pop; gt_as_hierfstat(): fix numeric labels
• gt_update_backingfile(): fix numbers of loci that are multiples of chunk_size; fix printing gen_tibbles with no loci
• fix documentation and error messages of several functions

tidypopgen 0.4.4

• fix seed argument of gt_admixture()
• implement our own upset plot function to avoid dependency on UpSetR package, which is no longer maintained
• fix reading of VCFs when genotype separator symbol is present in the sample information
• fix loci_ functions output when calculating across groups for a single loci
• fix gt_update_backingfile() to preserve imputation metadata when backing files are updated
• add validation in gt_impute_simple() to detect and report SNP dimension mismatches caused by excluded SNPs
• add outdir argument in gt_admixture() to store partial outputs for large jobs
• fix gt_admixture() output file names when n_runs > 1

tidypopgen 0.4.3

• fix occasional test failure when temporary file names contained certain patterns

tidypopgen 0.4.2

• allow least squares projection of multiple components at once in gt_project_pca()

tidypopgen 0.4.1

• duplicate-distance checks removed in is_loci_table_ordered()
• make qc_report_* functions work with pseudohaploids

tidypopgen 0.4.0

• store chromosomes as factors, and remove redundant column storing chromosome as integer

tidypopgen 0.3.3

• only use the FBM backend (rather than a complete bigsnpr object)
• fix a few minor bugs

tidypopgen 0.3.2

• tidy up for CRAN submission

tidypopgen 0.3.0

• add gt_pseudohaploid() and adapt appropriate functions to work with pseudohaploids

tidypopgen 0.2.0

• Lots of optimisations to speed up processing of large datasets.
• Added an sf backend to tidypopgen for spatial data.
• Windowed statistics.

tidypopgen 0.1.0

• Initial public version of tidypopgen released on r-universe.