NEWS.md

In OchoaLab/popkinsuppl: Supplement to popkin package

popkinsuppl 1.0.0.9000 (2019-05-22)

• First public version on GitHub!

popkinsuppl 1.0.1.9000 (2019-05-23)

• Changes shared by all FST methods (WC, Hudson K, and Hudson Pairwise).
• Improved speed and memory usage
• Added m option, required when genotypes are passed via a function (rare)
• Renamed option indexes_keep to ind_keep, corrected documentation to say it filters individuals (incorrectly said it filtered loci).

popkinsuppl 1.0.2.9000 (2019-05-30)

• Added kinship_std_limit

popkinsuppl 1.0.3.9000 (2019-07-04)

• Fixed error in kinship_std description of mean_of_ratios parameter. Now it correctly says that FALSE is the default value.

popkinsuppl 1.0.4.9000 (2019-07-24)

• Adapted package to use new solve_m_mem_lim from the popkin package, which generalizes some of the calculations that used to be part of the memory estimation/control code in get_mem_lim_m_kinship_std and get_mem_lim_m_WC.
• kinship_std now preserves individual names when present in input genotype matrix.

popkinsuppl 1.0.5.9000 (2019-08-02)

• Internal changes to match updated solve_m_mem_lim from the popkin package (>= 1.2.6.9000).

popkinsuppl 1.0.6.9000 (2019-12-17)

• Preemptively updated class usage now that matrices return a two-element array in R-devel
• Minor Roxygen-related updates

popkinsuppl 1.0.7.9000 (2020-01-29)

• Implemented the Weir-Hill 2002 FST estimator (fst_wh).

popkinsuppl 1.0.8.9000 (2020-05-07)

• fst_wc fixed a minor bug: sample size was half as it should have been, only noticeable in very small samples.
• Updates to fst_wc, fst_wh, fst_hudson_k, and fst_hudson_pairwise
• Avoids NAs in output when entire subpopulations have missing data at one locus
• Works correctly when there are singleton subpopulations (used to cause errors)
• More clear error messages for fewer than two subpopulations
• Expanded tests to cover these scenarios

popkinsuppl 1.0.9.9000 (2020-05-21)

• Reimplemented fst_hudson_pairwise to make it much more efficient in large datasets
• Original implementation simply called fst_hudson_k separately for every pair of subpopulations, which caused excessive and non-optimal file reading, which scaled increasingly poorly as the number of subpopulations increased.
• New implementation directly constructs pairwise FST matrix, reading input genotype matrix only once.
• The new implementation also better handles missing genotypes and is more numerically stable than before.
• In the public Pacific dataset (Skoglund et al. 2016) it was verified to agree with the Hudson implementation in EIGENSUITE 7.2.1 (executable smartpca)

popkinsuppl 1.0.10.9000 (2020-05-25)

• Reverted fst_wc sample size formula to what it was prior to version 1.0.8.9000. Bias tests showed that the original sample size equation was best.

popkinsuppl 1.0.11.9000 (2020-08-27)

• Function fst_wc has new option FIT, which if TRUE additionally includes calculations of the total inbreeding (FIT) next to the default structural inbreeding (FST) output list.

popkinsuppl 1.0.12.9000 (2020-08-30)

• Debugged functions fst_wc and fst_wh, which previously propagated some matrix-vector products incorrectly. This bug resulted in small biases when subpopulation sample sizes were unequal (sample weighing was set incorrectly; there was no bug when all sample sizes were equal). The bug was introduced in version 1.0.8.9000 (commit b445e3e) and remained until this version.

popkinsuppl 1.0.13.9000 (2020-09-24)

• Function kinship_std
• Added want_M option, which if TRUE returns a list containing the kinship matrix as well as the pairwise complete count matrix M.
• Added m_chunk_max option (default 1000), which sets the maximum number of loci to process at the time. The new default behavior reduces memory usage a lot, especially on machines with large memory, without sacrificing speed. Original version would use a lot of memory just because it was available, which could be inconvenient when trying to run other processes, and did not result in increased speed, so it was unnecessary at best.
• Fixed a rare bug that affected mean_of_ratios = FALSE version only, and only occured when m_chunk was smaller than m_loci. In original version m_chunk was as large as memory allowed, so the bug would have only occurred when m_loci was very large.

popkinsuppl 1.0.14.9000 (2020-10-16)

• Extended m_chunk_max option (see kinship_std above) to functions: fst_wc, fst_wh, fst_hudson_k, and fst_hudson_pairwise. This reduces memory usage a lot without sacrificing speed.

popkinsuppl 1.0.15.9000 (2021-02-23)

• Added function kinship_wg_limit, calculates the limit of the Weir-Goudet (WG) estimator.
• Also obtains WG kinship estimates if input is the popkin kinship estimate instead of the true kinship matrix, which is validated in internal tests against the WG implementation from SNPRelate.

popkinsuppl 1.0.16.9000 (2021-08-26)

• Added function kinship_gcta_limit.
• Updated DESCRIPTION main paragraph to mention kinship biased limit functions.
• Reformatted this NEWS.md slightly to improve its automatic parsing.
• Removed "LazyData: true" from DESCRIPTION (to avoid a new "NOTE" on CRAN).

popkinsuppl 1.0.17.9000 (2022-05-04)

• Added tests that confirm that bias transforming popkin estimates recovers standard kinship estimates when there's no missingness (for both ROM and MOR cases!)
• Added test_that section for comparisons to the SNPRelate package (mostly changed indentation).

popkinsuppl 1.0.18.9000 (2022-05-11)

• Updates centered around the fact that the GCTA GRM is the standard kinship estimator (MOR), has been for most of GCTA's existence.
• I had the impression before that GCTA used a different formula described in a 2011 paper, but it was abandoned in software in 2011 in favor of the standard estimator!
• Added a toy test that confirms that a precalculated GCTA GRM equals the output of kinship_std with mean_of_ratios = TRUE from the same input genotypes.
• Updated the documentation for kinship_gcta_limit to clarify my recent findings.
• Added function inbr_gcta. which I planned to validate against GCTA's estimate when I discovered my flawed assumption. I present it anyway since it was a lot of work, though it is a historical dead end most likely.
  • Function has internal validation (two formulas are implemented) and was validated against the limit kinship_gcta_limit in an admixture simulation (not part of this package).

OchoaLab/popkinsuppl documentation built on May 17, 2022, 9:50 a.m.