bug in relatedness matrix computation when parents are referenced but missing from the pedigree

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: BGmisc
 Title: An R Package for Extended Behavior Genetics Analysis
-Version: 1.8.0
+Version: 1.8.0.9
 Authors@R: c(
     person("S. Mason", "Garrison", , "garrissm@wfu.edu", role = c("aut", "cre"),
            comment = c(ORCID = "0000-0002-4804-6003")),

NAMESPACE

@@ -1,6 +1,12 @@
 # Generated by roxygen2: do not edit by hand
 
 export(SimPed)
+export(addMaternalChain)
+export(addMaternalLineFlag)
+export(addParentalChain)
+export(addParentalLineFlag)
+export(addPaternalChain)
+export(addPaternalLineFlag)
 export(addPersonToPed)
 export(alignPhenToMatrix)
 export(buildFamilyGroups)
@@ -73,6 +79,9 @@ export(vech)
 importFrom(Matrix,Diagonal)
 importFrom(Matrix,sparseMatrix)
 importFrom(data.table,as.data.table)
+importFrom(igraph,V)
+importFrom(igraph,distances)
+importFrom(igraph,subcomponent)
 importFrom(stats,na.omit)
 importFrom(stats,pnorm)
 importFrom(stats,qnorm)

NEWS.md

@@ -1,8 +1,17 @@
 # BGmisc NEWS
+
+## Development version
+
+* Add building of parental chains with `addParentalChain()` and `addParentalFlag()`, so that you can easily trace maternity and paternity. This is implemented as a general function that can be used to build any parental chain, and a specific wrapper for maternal and paternal chains. The parental flag is a binary variable that indicates whether the individual is in the specified parental chain, which can be useful for filtering or grouping individuals based on their lineage.
+* Fixed silent mis-scoring in `ped2com()`/`ped2add()` (and other component wrappers) when `momID`/`dadID` reference a parent ID that has no row of its own in `ped` (e.g., unrecorded founder stock, or a pedigree subset that excludes the parent's own row). Previously, `isChild_method = "partialparent"` treated such a parent as "known" (based on `momID`/`dadID` not being `NA`) while the adjacency builders correctly treated the link as absent, so the diagonal was silently understated (e.g., 0.5/0.75 instead of 1) and covariance between siblings who shared the missing parent was lost entirely. `ped2com()` now warns when this is detected. Two solutions are available in the new `repair_rowless_parents = TRUE` argument.
+    * \code{"rows"} (default) adds one placeholder founder row per unique missing parent ID (not one per affected child) to a working copy of \code{ped}, then restricts the returned matrix back to the original individuals via \code{keep_ids} unless \code{keep_ids} is already supplied. 
+    * \code{"schur"} applies a Schur complement on the block-triangular RAM system -- for each missing parent, its already-known children define a rank-1 update (\code{v \%*\% t(v)}, where \code{v} is that parent's traced genetic contribution to every individual, computed from the existing RAM matrix) which is added to the relatedness matrix at the tcrossprod step. \code{"schur"} currently only supports \code{component = "additive"}.
+
 ## BGmisc 1.8.0
 * Optimized gedcom reader, com2links for speed and memory usage, with a focus on large pedigrees
 * Fixed bug in gedcom reader that resulted in document records being added to the final person in the pedigree
 * Added more unit tests for gedcom reader and data parser
+* several improvements to the GEDCOM parsing functionality, focusing on more robust and flexible event parsing, better support for different GEDCOM versions, and enhanced usability.
 * Optimized sliceFamilies to be more abstract, and no longer require mtdna
 * Created `.require_openmx()` to make it easier to use OpenMx functions without making OpenMx a dependency
 * Smarter string ID handling for ped2id