PedCompare: When dummy parent mismatched 1 or more
dummy offspring than it had correct genotyped offspring, all genotyped +
dummy offspring were set to ‘nomatch’. Now match is kept, and only
actual mismatched flagged as such.ComparePairs: New argument ‘Pairs_suffix’, passed to
getRelM.getRelM: when only Pairs is provided,
default (for unlisted pairs) changed from ‘U’ to ‘X’SimGeno, throwing an error when ParMis=c(0,
not-zero)CalcPairLL: force to (almost) always calculate LL even
when highly unlikely, by fixing MaxMismatch to the total number of SNPs
and T_filter to -999.0.CalcParentProbs: new function to calculate assignment
probabilities for any pedigreegetAssignCat now also considers parents with only
dummyfiable offspring as dummyfiable; option ‘1sib’ is dropped.PedCompare: when an incorrect half-sibling is added to
a true singleton-sibship-with-grandparent, this now counts as 1
mismatch; this erroneously was 3 mismatches (entire sibship wrong).sequoia, the default value for
CalcParentLLR is changed from TRUE to FALSE.Err_Allele2LocusModule='ped'GenoConvert fake .map output file: change chrom 0
(unmapped) to chrom 1, as SNPs on chromosome 0 get excluded by default
by e.g. PLINK & GCTACalcRped: drop dummy parents of ‘half founders’ from
output (1 known + 1 unknown parent not supported by pkg kinship2)GenoConvert: new vcf and genlight input,
various bugs fixed, behaviour made more consistent and clearer through
additional messages.GenoConvert and
SnpStatsMkGenoErrors (used by SimGeno)
causing about 3x too many hom|hom errors when SnpError is a
single value: first beta-distributed per-SNP genotyping error rates
\(E_l\) were generated, and then \((E_l/2)^2\) calculated. Now the single
value is by default first morphed into a length 3 vector (hom|hom,
het|hom, hom|het), and three beta distributions are generated.MkGenoErrorsSimGeno ParMis default changed from 0.4 to 0ErrToM).EstEr (estimation of genotyping errors)
removed due to inaccurate estimations and misuse. Will (probably) be
re-implemented in a future version.CalcMaxMismatch: OH with both parents counts
as 2 mismatches (was 1)GenoConvert, and fix bug with
OutFormat ‘col’EstConf example to nSim=1 to ensure runtime <
5 sec to pass CRAN checkmtSame: specify if individuals have the same or
different mitochondrial haplotypeCalcPairLL helpfileCalcBYprobs, which caused Year.last to be
ignoredLifeHistData$Sex includes
NAYear.last added to
LifeHistData (last possible offspring birth year)GetAncestors and
GetDescendantsMinAgeParent for
MakeAgePrior()StrictGenoCheck, Strict for
CheckGeno() ; update msgsCheckLH, now flexible column order in
LifeHistData.qntl = 0.999^(1/nrow(GenoM))
to 0.9999^(1/nrow(GenoM)) in all functions calling
CalcMaxMismatchSnpStats() when AF=0 or SNP is missing for
all individuals.sequoia_report()PedCompare() parameter minSibSize (minimum
sibship size for which the non-genotyped parent is considered
‘dummyfiable’) changed default value from 2sib (2 genotyped
siblings) to 1sib1GP. This reflects the increased success
of reconstructing grandoffspring-grandparent pairs in newest version,
which make the output of PedCompare(,minSibSize='2sib')
confusing. This also affects EstConf().GetRelM() now allows input of Pedigree AND Pairs. This
a.o. allows PlotRelPairs() with both inferred pedigree plus
GetMaybeRel() output.writeSeq()
(easier to install)SeqOUT_HSg5 for quicker vignette compilationGeno_griffin,
Conf_griffin (output from EstConf()) and
MaybeRel_griffin (output from GetMaybeRel()),
as well as the script used to create these
(mk_griffin_data.R)SummarySeq() was internal and
is now exported (PlotSeqSum())sequoia() now includes
args.AP.SummarySeq(), if Pedigree is provided
rather than SeqList and SNPd=NULL, all
individuals are categorised as Observed (was:
Genotyped).PedPolish() now has arguments to specify whether to
drop extra columns (besides id-dam-sire) and whether to keep rows with
non-unique or NA ids.SnpStats() now includes HWE testsSimGeno() for non-autosomal
inheritance; this option is still experimental and non-autosomal SNPs
are not supported by the pedigree reconstruction.cannot allocate vector of size ... issue in
GetRelM() with very large pedigrees, which affected
MakeAgePrior() and thereby sequoia().EstConf() $ConfProb used the wrong
denominator, namely the number of parents in the reference pedigree
rather than in the inferred pedigree.CalcPairLL(), as
LL(HS)=LL(GP)=LL(FA) for these pairs. Origin time unknown.GenoConvert() when
Informat='single'.CalcPairLL() HS likelihood when
conditioning on pedigree was incorrect. No/minimal effect on pedigree
reconstruction.DuplicateCheck() (always automatically
called by sequoia()) that on very rare occasions caused R
to crashfixes minor bugs identified by CRAN valgrind and gcc-ASAN
minor edits to vignette to comply with CRAN precheck
CalcBYprobs() to estimate the probability
that individual i is born in year y.EstConf() outputGenoConvert() regarding InData vs InFileErr=0Herm instead of specified via
Complex. New output list element
DummyClonesCalcRped() to calculate pedigree
relatedness. Uses package kinship2.Various smaller bugs have been fixed, some affecting assignment rate
sequoia(): Option FindMaybeRel has been
deprecated; call GetMaybeRel() directly instead.CalcPairLL(), returns likelihoods for each
of the 7 considered relationships (PO, FS, HS, GP, FA, HA, U) for each
specified pair of individualsRelPlot() for Colony-like visualisation of
pairwise relationships (automatically called by
ComparePairs())sequoia(): fixed error
object 'ErrM' not found when re-using output from a
previous sequoia run. Circumvent this bug in version 2.0.7 by fooling
the program to think it’s output from an older version:
names(ParOUT$Specs)[match("MaxMismatchOH", names(ParOUT$Specs))] <- "foo".sequoia(): fixed bug causing genotyped parents to not
always be monogamous when Complx='mono'.Error in data.frame(id = c(dID[s(nd[1]), 1], dID[s(nd[2]), 2]), VtoM(TMP$dumparrf, : arguments imply differing number of rows)
)GetMaybeRel(): fixed error
(subscript) logical subscript too long when input pedigree
contains dummiesGetMaybeRel(): fixed error causing likely GP pairs not
to be included outputPedCompare(): fixed id.dam.cat and id.sire.cat being
‘NANA’ instead of XD, XG or XX when Symmetrical=TRUEPlotAgePrior(): Avoid using
grDevices::hcl.colors() in R versions <3.6, where this
function is not yet availableComparePairs(): fixed bug when Pairs2 but
not Ped2 is specifiedMakeAgePrior(): fixed bug when there are no FS pairs in
the input pedigreeMakeAgePrior(): MaxAgeParent was ignored
when a pedigree with overlapping generations was supplied\donttest instead of
\dontrun, so that they can be run using
example(). Note that some can be quite time consuming,
especially EstConf().PlotPedComp() to visualise
PedCompare() outputSimGeno(): deprecated input parameters (since v 1.3.1)
are dropped completelygetAssignCat() no longer drops additional columns from
input pedigreeCalcOHLLR(, CalcLLR = FALSE)PedCompare() output element DummyMatch now
also include output class of matched individual’s parents &
offspringsequoia() now only returns pairs for
which LL_duplicate - max(LL_{not duplicate}) > T_filter; when call
rates are low this may be a substantially shorter list than in previous
versions, where all pairs with fewer than MaxMismatchDUP differences
were listed.ComparePairs() can now be called for a single pedigree,
as well as to compare two pedigreesPlot=TRUE, all functions will print a message and return
results as usual, instead of throwing an error and not returning
results.MakeAgePrior(): more consistent implementation; is now
called by sequoia() with only lifehistory data of genotyped
individuals.EstConf() now also returns the full Counts
table from PedCompare(); $RunParams now holds
the evaluated input paramters, instead of e.g. V[i] when
called from inside a loop.ErrToM. The
default has changed very slightly from version 1.3.CalcOHLLR() to calculate Mendelian errors
and parental log-likelihood ratios for any pedigreegetAssignable() to flag genotyped and
‘dummifiable’ individuals in any pedigreeComparePairs() to compare pairwise
relationships between 2 pedigrees; replaces now-deprecated
DyadCompare.PedPolish() is now user available.MaxMismatch of function
sequoia, now calculated internally by new function
CalcMaxMismatch based on number of SNPs, presumed
genotyping error rate, and minor allele frequenciesEstConf() now also estimates confidence for
parent-pairs; output has changed considerably.PedCompare() to increase clarity of
code for easier maintenance; changed output format somewhat.GenoConvert() skipped first individual when reading
.raw file. Circumvent this bug in earlier versions by using option
header=FALSE (then header row is removed only once…)ConfProb() expected input parameter nSim
to be strictly integer, now relaxed to any value convertible to a whole
numberERROR! ***Invalid ParProb!*** triggered when some
SNPs are monomorphicSEGFAULT triggered when some SNPs have very high
missingness (>80%); possibly sibship size out of boundsError arguments imply differing number of rows
when there are dummy parents of 1 sex onlyLifeHistData may have 2 additional columns, with
minimum and maximum possible birth yearPed_griffin) to illustrate
overlapping generations, used in age vignetteSummarySeq(): added a pedigree summary table identical
to a subset of the table returned by R package pedantics’
pedStatSummary; that package has been archived on CRAN.
Added option Panels to only plot (a) specific
panel(s).CheckGeno(), MkGenoErrors(),
GetMaybeRel(), GetRelCat()PlotAgePrior() and
SummarySeq()SimGeno() input parameters have changed, old
ones will be deprecatedEstConf, SnpStatsFirst version on CRAN!