ChangeLog

			CHANGES IN ADEGENET VERSION 2.1.10

CRAN MAINTENANCE

- one final lingering short bool definition was purged

			CHANGES IN ADEGENET VERSION 2.1.9

CRAN MAINTENANCE

 - failures on C23 were fixed. The reserved word `bool` definition was
 replaced with including stdbool

			CHANGES IN ADEGENET VERSION 2.1.8

CRAN MAINTENANCE

 - Some testing code in C that was triggering warnings in clang was removed

			CHANGES IN ADEGENET VERSION 2.1.7

BUG FIX

 - as.matrix and as.data.frame methods for genind/genpop objects will now use
   `tab()` internally to avoid https://github.com/thibautjombart/adegenet/issues/331

DEPENDENCIES

 - Suggested packages tripack and akima have been removed and replaced with
   interp, which is used in the tutorials.
   (see https://github.com/thibautjombart/adegenet/issues/332)

			CHANGES IN ADEGENET VERSION 2.1.6

MISC

 - in `xvalDapc()`, if `n.pca.max` is a vector greter than length 1, the
   maximum value will be used (@pdimens, #323)

BUG FIX

 - `summary.spca()` now works when matricies and data frames are used as the
   input for `spca` (@lvclark, #329)


			CHANGES IN ADEGENET VERSION 2.1.5

PARALLEL COMPUTATION

 - parallel computation defaults to `FALSE` from now on to avoid frustrating
   errors on windows machines (@courtiol, #313).

DEPENDENCIS

 - spdep and adespatial have been moved to Suggests to make installation
   easier for people who do not need spatial functions.

BUG FIX

 - tab() method for genpop objects no longer return NaN for freq = TRUE when a
   locus is missing from a population (this affects hybridize; @Sergi-Taboada, 
   @eachambers, #262)

MISC

 - Mis-matched documentation in glPca() has been fixed (@plantarum, #298)

			CHANGES IN ADEGENET VERSION 2.1.4

DOCUMENTATION

  - `snapclust()` documentation has been updated to be not so scary
	(@thibautjombart, #287)

			CHANGES IN ADEGENET VERSION 2.1.3

BUG FIX

  - `seppop()` correctly subsets genlight objects now. (@rdinnager, #270)
  
MISC

  - A bug in the `scaleGen()` example has been fixed and a minor spelling
    mistake was corrected (@courtiol, #273)
	- `find.clusters()` gains a missing parenthesis in the user prompt 
  	(@knausb, #283)
	- Warning messages for missing loci and individuals have been made clearer
	  (@zkamvar, #276)
	- DESCRIPTION file now uses Authors@R syntax

CRAN MAINTENANCE

  - `DNAbin2genind()` now checks for matrices by using `inherits()` instead of
	  `class(x) == "matrix"` (@zkamvar, #275)

			CHANGES IN ADEGENET VERSION 2.1.2

NEW MANAGEMENT

  - Zhian N. Kamvar is now the acting maintainer of the package.

NEW FEATURES

  - While creating genind from dataframe, it is now possible to skip
    the test of the ploidy that was added in 2.1.1 by passing the option
    check.ploidy = FALSE to the function df2genind. Thanks to Frederic Michaud
    for the fix. See https://github.com/thibautjombart/adegenet/issues/231 and
    https://github.com/thibautjombart/adegenet/pull/232 for details

  - `showmekittens()` has gained several new kittens thanks to @rdecamp

  - A new data set called `swallowtails` has been added to demonstrate the
    export to MVmapper functionality.
    
BUG FIXES

  - A bug where the number of alleles per locus was unaffected after subsetting
    a population with `drop = FALSE` has been fixed. Now, by default, the 
    `@loc.n.all` slot (accessor: `nAll()`) will store the number of observed
    alleles in the current data set, regardless of whether or not these were
    dropped from the data matrix. 
    See https://github.com/thibautjombart/adegenet/issues/234 for details.
    
  - `seppop()` will now take into account `NA` population assignments with the
    new `keepNA` argument. 
    See https://github.com/thibautjombart/adegenet/pull/236 for details.

  - `snapclust()` likelihood calculation was incorrect for haploids. This has
    been fixed in commits 3a5f820db388c9821c3095cd29ad3cd3843237a5 and
    e41d2d598369bb5921465a60adffac9ecb61f86d. 

MISC

  - `snapclust()` will give a better warning message when the number of
    samples in the dataset exceeds computer precision. See
    https://github.com/thibautjombart/adegenet/issues/221 for details.
    
  - `df2genind()` now imports data >100x faster 
    (@KlausVigo, https://github.com/thibautjombart/adegenet/pulls/260)

			CHANGES IN ADEGENET VERSION 2.1.1
BUG FIXES

  - A warning will be issued if the user specifies a ploidy range that is less
    than the observed allele dosage when data is imported by df2genind. See
    https://github.com/thibautjombart/adegenet/issues/216 for details.

			CHANGES IN ADEGENET VERSION 2.1.0
NEW FEATURES
	- new set of functions implementing 'snapclust' for genetic
	clustering using fast maximum-likelihood estimation; new functions
	include: `snapclust`, `snapclust.choose.k`, and methods for `AIC`,
	`AICc`, `BIC`, `KIC`

	- compoplot is now a generic with methods for different objects,
	including dapc and snapclust

	- spca is now a generic with methods for genind, data.frame and
	genlight objects

	- a new generic wrapper `export_to_mvmapper` has been added, which
	exports multivariate analyses (e.g. DAPC, sPCA, or standard dudi
	objects) with geo-referenced units to serve as input to mvmapper
	(https://popphylotools.github.io/mvMapper/)

	- the new function `spca_randtest` implements a Monte Carlo test
	for spatial structure in the sPCA framework.

BUG FIXES
	- fixed registration of C routines which could cause problems with
	SNPbin and genlight objects.

	- fixed issues relating to `read.PLINK` and the reading of `.map`
	files (issues 94 and 188)

	- fixed issues relating xvalDapc with non-table objects by registering
	methods for genind and genlight object (issue 193)

	- find.clusters with clust parameter will no longer throw an error
	when attempting to find clusters of populations with fewer than
	n/10 individuals (issue 184)



			CHANGES IN ADEGENET VERSION 2.0.2
BUG FIXES
	o df2genind will now replace "." in alleles with "_" (or "p" if sep = "_")
	see https://github.com/thibautjombart/adegenet/issues/132 for details

	o if several locus names would partially match through grep,
	df2genind would output an incorrect genind object
	NA-wise. Reported by Elizabeth, see
	https://github.com/thibautjombart/adegenet/issues/160

			CHANGES IN ADEGENET VERSION 2.0.1
NEW FEATURES
	o Hs is now much faster and will scale better for large number of
	loci

	o Hs.test allows for testing differences in expected
	heterozygosity between two groups of individuals

	o strata tutorial is now available via
	adegenetTutorial(which="strata")

	o repool can now handle objects with no population information,
	and can return a list of repooled genind (argument 'list'
	defauling to FALSE)

	o read.fstat can now handle missing data coded by any number of
	"0"
	
	o `scatter.dapc` function can now label individuals using `orditorp`
	function from `vegan` package, which takes care of overlapping
	(See issue #100)
	
	o df2genind is now faster at handling missing data (See issue #114)
	
	o the summary method has better formatting

BUG FIXES
	o Hs no longer returns NaN if one locus has not been typed;
	instead, computations are done using all available loci

	o adegenetTutorial now opens up-to-date tutorials

	o subsetting genlight objects now treats missing data appropriately
	when given logical, character, or negative subscripts.
	(See issue #83 (https://github.com/thibautjombart/adegenet/issues/83))

	o fixed occasional spurious warning about duplicated individual
	labels in df2genind
	
	o fixed issues with mis-placed missing data in df2genind when samples
	were labeled with numbers (See issue #96)
	
	o frequencies can now be obtained from P/A genpop objects (See issue #105)
	
	o Windows no longer throws an error with as.genlight (See issue #109)
	
	o read.genpop now returns individual names (See issue #117)



			CHANGES IN ADEGENET VERSION 2.0.0
CHANGES IN GENIND/GENPOP/GENLIGHT OBJECTS
	o generic labels are no longer used; therefore, 'truenames' is now
	deprecated

	o @tab now stores integers representing counts of alleles, as
	opposed to the former numeric for allele frequencies

	o @ploidy is now a vector with one value per individual; ploidy
	can differ across individuals

	o @loc.names, @ind.names, @pop.names have been removed

	o accessors locNames, indNames and popNames can be used to replace
	calls to the removed slots

	o hierarchical structure is implemented in genind and genlight
	objects, by the addition of two new slots @strata and @hierarchy.

	o subsetting genlight objects is now up to 10x faster.


NEW FEATURES
	o ploidy can now vary across individuals

	o printing is now consistent across objects classes

	o summary methods now get a verbose argument

	o df2genind is more flexible and faster

	o genind objects size is roughly divided by 2

	o adegenet is now better integrated alongside other packages
	including pegas and hierfstat; functions for HW test and F
	statistics have been migrated into these packages and removed from
	adegenet or deprecated

	o new accessor tab(...) replaces 'truenames' and should be used to
	get tables of allele counts, frequencies, and optionally replace
	missing data

	o new accessor locNames<- can be used to rename loci in
	genind/genpop

	o new accessors 'strata' and 'hier' can be used to get/set
	hierarchical clustering information!

	o genind2genpop can now take a formula argument for pop to
	specify strata.

	o subsetting of genind/genpop is more powerful, allowing:
	x[loc=a , pop=b] where a and b are any subset of loci and
	populations (characters, integers, logicals)

	o dapc can now store loadings of original variables (argument
	var.loadings=TRUE)

	o hybridize is now more tolerant towards heterogeneous parent
	populations and repools data systematically

	o xvalDapc can now run in parallel!

	o roxygen2 is now used for part of the documentation, and
	mandatory for new contributions

	o new tutorials (distributed on github and on the adegenet
	website) are now available, including one on strata

	o adegenetIssues() will take you to the issues page of the project
	on github

	o new welcoming message upon loading the package!

	o a hidden function for listening to music



			CHANGES IN ADEGENET VERSION 1.4-3
NEW FEATURES
	o the generic function pairDist has been added; it plots pairwise
	distance distributions for different group combinations. Method
	for objects of class: dist, matrix, genind, DNAbin.

	o 'inbreeding' can now return the maximum likelihood estimate via the
	stats function 'optimize'.



			CHANGES IN ADEGENET VERSION 1.4-2
FIXES
	o fixed adegenetServer; this should now work on any platform.



			CHANGES IN ADEGENET VERSION 1.4-1
NEW FEATURES
	o the adegenet website is now indicated in CRAN

FIXES
	o fix to adegenetTutorial: the basics tutorial would not be opened
	by the function



			CHANGES IN ADEGENET VERSION 1.4-0
NEW FEATURES
	o adegenet now incorporates a web interface for DAPC, started by
	the command adegenetServer("DAPC")

	o 'xvalDapc' has been modified to incorporate heterogeneous sample
	sizes

	o the new function 'snpzip' implements feature selection using DAPC

	o 'snpposi.plot' plots SNP density across a DNA sequence

	o 'snpposi.test' tests for randomness in the distribution of SNPs
	across a DNA sequence

	o the package is no longer distributed with vignettes; instead,
	tutorials are available from the adegenet website, and can be
	accessed from R directly by the function 'adegenetTutorial'.



			CHANGES IN ADEGENET VERSION 1.3-9.2
FIXES
	o fixed NAMESPACE to get rid of a warning due to conflicting
	imports in ape and igraph

	o fixed seploc and seppop, which were loosing some of the
	attributes of the objects (ploidy, type)


			CHANGES IN ADEGENET VERSION 1.3-8

NEW FEATURES

	o new palettes: azur, wasp

	o new function any2col translates (numeric, factor, character)
	vectors into colors, also providing information for a legend

	o new function xvalDapc which performs cross-validation for a dapc
	analysis.


			CHANGES IN ADEGENET VERSION 1.3-7

NEW FEATURES

	o entirely new version of DNAbin2genind, much faster than before
	and suitable for large alignments



			CHANGES IN ADEGENET VERSION 1.3-6

NEW FEATURES

	o findMutations: a new procedure to identify the location and
	nature of mutations between pairs of DNA sequences

	o graphMutations: a graphical representation for findMutations

	o improved graphics for gengraph



			CHANGES IN ADEGENET VERSION 1.3-5

NEW FEATURES

	o seqTrack and haploGen now have export functions to igraph class.

	o seqTrack and haploGen now have default plot methods relying on
	igraph conversion.

	o fstat and gstat.randtest have been restored.

	o gengraph implements graph-based approaches for representing
	genetic diversity (e-burst type of approaches for any time of
	genetic data).


BUG FIXES

	o mutation rates have been fixed in haploGen

	o calls to printf replaced by Rprintf in C procedures

	o seqTrack example fixed (conversion to graphNEL removed, now
	using igraph)

	o DLL is now loaded within the NAMESPACE, .First.lib is no longer
	used.



			CHANGES IN ADEGENET VERSION 1.3-3

BUG FIXES

	o fixed a bug of propShared, which gave wrong results under weird
	circumstances. The new implementation is entirely different, uses
	C code, and is now applicable to data with any level of ploidy.

	o tried making the package smaller by removing unnecessary files.




			CHANGES IN ADEGENET VERSION 1.3-2

BUG FIXES

	o replaced calls to multicore:::detectCores with
	parallel:::detectCores (the former gives unexpected results on
	some platforms).

	o documentation update: fstat function is no longer available due
	to removal of hierfstat package from CRAN. An example shows how to
	use Fst function from the pegas package instead.

	o documentation update: doc now specifies that read.genepop and
	read.fstat are meant for diploid data only, with guidelines for
	haploid data.



			CHANGES IN ADEGENET VERSION 1.3-1

BUG FIXES

	o removed all dependency with graph package, which was still
	causing errors on some systems - whenever Bioconductor was not by
	default one of the accessible repositories (used to be FAQ #4).

NEW FEATURES

	o legend in scatter.dapc now matches to the type and size of
	symbols used in the plot.

	o the package has been made smaller (from 20MB to 8.5MB).

	o a warning has been added to read.structure: the function is made
	for diploid data only.



			CHANGES IN ADEGENET VERSION 1.3-0

BUG FIXES

	o fixed broken dependencies with hierfstat and graph packages
	(used to be FAQ #4).

	o fixed a minor bug in loadingplot regarding possible label errors.

NEW FEATURES

	o genome-wide SNP data support using the new class genlight,
	supported by compiled C routines and parallelized computations on
	multicore architectures.

	o dedicated find.clusters and dapc methods for genlight objects.

	o read.PLINK to read SNP data with PLINK format into genlight
	objects.

	o read.snp to read SNP data with adegenet's own format into genlight
	objects.

	o fasta2genlight to extract SNP from FASTA files into genlight
	objects.

	o new method 'predict' for DAPC objects, allowing for using
	supplementary individuals.

	o many new options of DAPC scatterplots (scatter.dapc).

	o new plotting method 'compoplot' for DAPC objects, displaying
	group memberships in a STRUCTURE-like way... only prettier.

	o many new accessors for the classes genind and genpop.

	o 4 new tutorial vignettes: adegenet-basics, adegenet-dapc,
	adegenet-spca, adegenet-genomics.

	o last but not least: Ismail Ahmed has joined the project as a
	developper.



			CHANGES IN ADEGENET VERSION 1.2-8

BUG FIXES

	o fixed a major issue in the conversion of genind objects to
	hierfstat data format. In some cases, this issue biased the
	results of the wrapper function "fstat".



			CHANGES IN ADEGENET VERSION 1.2-7


NEW FEATURES

	o dapc and find.clusters are now stable versions for the published methods.

	o a.score and optim.a.score are released in their beta versions.

	o scatter.dapc is now adapted to representation 1-dimensional DAPC results.


BUG FIXES

	o fixed a major issue in the ape package causing haploGen to
	bug. This is a temporary replacement, waiting for ape to implement
	the changes (currently, adegenet replaces ape's as.list.DNAbin
	function with a corrected version).



			CHANGES IN ADEGENET VERSION 1.2-6


NEW FEATURES

	o pairwise.fst: computes Nei's pairwise Fst between populations

	o alignment2genind: extract polymorphism from nucleic and proteic
	aligned sequences with the 'alignment' format, returning a genind object.


BUG FIXES

	o fixed a minor issue in Hs, occuring in fixed loci for a given population.



			CHANGES IN ADEGENET VERSION 1.2-5


NEW FEATURES

	o SeqTrack, an algorithm for the reconstruction of genealogies, is
	now fully implemented and documented.

	o haploGen, a system for simulating genealogies, is now fully
	implemented and documented.


BUG FIXES

	o summary methods for genind and genpop are fixed (again) for the
	new R version 2.11.1.



			CHANGES IN ADEGENET VERSION 1.2-4


NEW FEATURES

	o Hs computes the theoretical heterozygosity by populations for
	genpop object

	o propShared is now available for haploid data

	o Discriminant Analysis of Principal Component (DAPC) is
	implemented by the function DAPC, although the method itself is
	still under review.

	o The SeqTrack algorithm for reconstructing genealogies is
	implemented by the function seqTrack, although the method itself
	is still under review.

	o Pre-release of new simulation tools, still undocumented
	(haploPop and haploGen).

	o New datasets eHGDP and H3N2


BUG FIXES

	o minor bug fixes in df2genind, propShared, and seploc

	o summary issue arised in certain conditions, depending on which
	package was loaded before adegenet; in some cases, the summary
	procedure was not found for genind/genpop objects. Fixed now.



			CHANGES IN ADEGENET VERSION 1.2-3


NEW FEATURES

	o implement handling of presence/absence markers. genind and
	genpop object now have a 'type' attribute to differentiate between
	codominant markers (e.g. microsatellite), which is the default
	type, and presence/absence data (e.g. AFLP). Functions in adegenet
	now behave according to the type of markers being used.

	o SNP can now be obtained from sequence data, stored as DNAbin
	(see E. Paradis's package 'ape'). They can be selected to verify
	any given degree of polymorphism.

	o 'hybridize' can now be used for genotypes having any even degree
	of ploidy (e.g. tetraploid genotypes).

	o the new function 'isPoly' checks which loci are polymorphic, or
	which alleles contribute to polymorphism.

	o the new function 'pop' can be used to retrieve and set the pop
	slot of genind object.

	o the new function 'selPopSize' allows one to select a subset of
	genotypes belonging to well-sampled populations, as defined by a
	threshold sample size.

	o the new accessor 'locNames' can be used to retrieve real labels
	of markers and/or alleles.

	o the loadingplot has been modified to allow specifying x axis, so
	that scoring SNPs along their sequence is now possible.


BUG FIXES

	o no bug to fix this version!



			CHANGES IN ADEGENET VERSION 1.2-2


NEW FEATURES

	o implement different levels of ploidy in genind / genpop objects
	(new slot @ploidy). Now, any level of ploidy can be handled by
	input function df2genind, which has been recoded almost
	entierely. Different levels of ploidy are now handled by different
	functions.

	o a "sep" argument is now handled by df2genind: this allows
	reading many data formats.

	o implemented a method "scaleGen" for genind / genpop objects,
	which scales allelic data using different methods.

	o  colorplot: a generic function, with a method for spca
	objects. Represents up to three principal components based on RGB
	representation of Cavalli-Sforza.

	o loadingplot for plotting loadings of one axis

	o adegenetTutorial function which opens the online tutorials

	o allow for the use of na.replace and scaleGen in spca function

	o added rupica dataset

	o enable reading data from URL (import2genind, read.[...])

	o permit specification of a matrix of spatial weights in spca


BUG FIXES

	o fixed bug 1.2-2.01 (read.structure issue): was due to the
	default of "onerowperind" argument.

	o fixed bug 1.2-2.02 (read.genetix issue): was due to an
	error in the data file (wrong nloc); now read.genetix corrects
	that automatically and issues a warning.

	o fixed bug 1.2-2.03 (monmonier issue): was a non-detected
	code 2 due to intersection check with previously drawn segment
	(was not always removed).

	o fixed bug 1.2-2.05 (a few fixes/improvement for monmonier)



			CHANGES IN ADEGENET VERSION 1.2-1


NEW FEATURES

	o documentation of scaleGen provides an example of usefulness of
	an appropriate scaling in PCA

BUG FIXES

	o fixed the recognition of NAs in df2genind

	o fixed the call to inherits in spca (returned value changes in R-devel)



			CHANGES IN ADEGENET VERSION 1.2-0


NEW FEATURES

	o implement different levels of ploidy in genind / genpop
	objects. Make necessary adaptations throughout the package.

	o put some stop where needed when ploidy!=2 is not handled.

	o implement a "sep" argument in df2genind.

	o implement accessor for genind/genpop: nLoc.

	o implement "scaleGen" for genind/genpop, which allows for
	different types of scaling.

	o added several coercion methods, from genind/genpop to
	data.frame, matrix and ktab objects.

	o implemented propTyped, a function giving the proportion of
	non-missing data in different ways.

BUG FIXES

	o missing data indicated in summary corrected (loci with more
	alleles had more weight in the computations).




			CHANGES IN ADEGENET VERSION 1.1-3


NEW FEATURES

	o 'as' methods for genind/genpop objects to matrix, data.frame,
	and ktab objects. Now, ordination implemented as dudi functions in
	ade4 (like dudi.pca) can be performed directly using genind/genpop
	as inputs.




			CHANGES IN ADEGENET VERSION 1.1-2


NEW FEATURES

	o significant improvement in the speed of genind2df (more than
	twice as fast as before).

	o function propShared added: computes the proportion of shared
	alleles among a set of genotypes (core computations in C).

	o A warning is issued when NAs exist in the input of sPCA.

	o improvement of the validity checking for genind/genpop:
	validObject now detects duplicates in any kind of names (ind.names,
	pop.names, etc.) and prints the corresponding items.

BUG FIXES

	o genind2df does now handles the pop argument correctly.

	o df2genind does no longer bug when there is an entirely non-typed
	locus.



			CHANGES IN ADEGENET VERSION 1.1-1


NEW FEATURES

	o I/O: df2genind no longer fails when entirely non-type
	individuals exist.

	o Monmonier: optimize.monmonier now computes the 'best'
	boundary only once instead of twice. The whole code was re-thought
	and optimized for speed. Monmonier's boundaries can now form
	loops. Instead of stoping at a given threshold, it is also
	possible to ask for a given length of boundary (argument
	bd.length).

	o The function chooseCN has a new option to return a list of
	spatial weights defined as the inverse of spatial distances, at a
	given exponent.

	o A wrapper for glob.varcomp has been implemented for genind
	objects, through the new function fstat.

	o The elements of the @other slot are now proceeded wisely when
	objects are subsetted using the '[' operator.


BUG FIXES

	o I/O: df2genind no longer fails when entirely non-type
	individuals exist.

	o monmonier no longer fails when coordinates are drawn from a
	regular grid. The matched call of the returned object has been
	fixed.



			CHANGES IN ADEGENET VERSION 1.1-0

NEW FEATURES
	o Data representation: S4 classes in replacement of old S3
	classes.

	o Spatial genetics: the spatial Principal Component Analysis
	(Jombart et al, 2008, Heredity), two multivariate spatial
	tests, and new functionalities for Monmonier's algorithm.

	o I/O: functions to import data are now 'read' functions;
	available for formats of GENETIX, Fstat, Genepop, STRUCTURE and
	from data.frames of genotypes. Export from genind to data.frame of
	genotypes.

	o Data: five new simulated geo-referenced datasets

	o Simulations: a hybridize function, which creates hybrids from
	two parent datasets. Can output to STRUCTURE format.

	o Data manipulation: new function to separate data by
	population. Accessors to genind and genpop object like with
	matrices using 'foo[ chosenGenotypes, chosenAlleles]'.



			CHANGES IN ADEGENET VERSION 1.0-2

NEW FEATURES

	o adegenetWeb is a simple function opening the adegenet website in
	the default web browser.

	o sim2pop is a dataset obtained by simulation using the software
	Easypop. It contains 130 georeferenced genotypes sampled from two
	distinct populations.

	o monmonier documentation was improved by adding a genetic
	example, using sim2pop data.

BUG FIXES

	o some bugs corrected in optimize.monmonier


			CHANGES IN ADEGENET VERSION 1.0-1

NEW FEATURES

	o chooseCN is a simple interactive tool for choosing and building
	a connection network from spatial coordinates. This tool is called
	by monmonier function.

	o monmonier, optimize.monmonier, plot.monmonier and print.monmonier
	implement the Monmonier algorithm. While not restrained to genetic
	data analysis, this method can be used to find genetic boundaries
	among individuals or populations based on their allelic
	frequencies and spatial coordinates.

BUG FIXES

	o several bugs fixed in I/O functions