Changed read_csv() to fread_csv() to avoid conflict with readr::read_csv(). (Issue #223)
Similarly, changed read_csv_numer() to fread_csv_numer().
Added function fund_dup_markers(), for identifying subsets of markers with identical genotype data. This is a port of qtl::findDupMarkers().
Have calc_het() stop with an error if the genotypes have labels that aren’t two characters, and add an explanation of this in the help info. (Issue #220)
More fully explain the use of weights in est_herit() and scan1(). (Issue #221)
create_variant_query_func(), added new arguments id_field and sdp_field, and in create_gene_query_func(), added arguments name_field and strand_field (Issue #215). This gives new flexibility, but also adds new requirements (for example, that the variant database has a field "snp_id") and so could potentially break working code.smooth_gmap() for smoothing out a genetic map, particularly to eliminate intervals with 0 recombination, by using a “mixture” of the map and constant recombination. Also added unsmooth_gmap() which does the reverse.read_cross2() now gives a warning if sex isn’t provided but is needed. Also, if sex is missing we assume all individuals are female; previously we assumed they were male. (Issue #214)
In plot_genes(), allow strand to be +/- 1 and not just "+" or "-". (Issue #216)
Fixed date in citation, Broman et al. (2019) doi:10.1534/genetics.118.301595
plot_genes() there was a case where stop was used that should have been end.calc_genoprob() for the X chromosome, so that it just keeps track of the chromosome from the DO/HS parent. In males, we are assuming that the DO/HS parent is the mother.Added dependency on version of Rcpp (>= 1.0.7)
Revised genoprob_to_alleleprob() to work with DOF1 and HSF1. plot_onegeno() should also work now in these cases. (Issue #140 and Issue #141)
Revised predict_snpgeno() to work for DOF1 and HSF1 populations.
Now give a better error message in genoprob_to_snpprob() if snpinfo is missing the sdp column (Issue #207).
In read_csv(), now give warnings if there are duplicate column names or duplicate row names in the file.
In read_cross2(), moved the warning regarding the number of alleles to before the alleles object gets corrected (Issue #209).
Now issue a warning message if founder genotypes are included but not used (Issue #211).
The treatment of the male X chromosome in DOF1 and HSF1 was incorrect. We’re now assuming that the DO or HS parent was the mother in the F1 cross, in which case males will be hemizygous for one of the DO/HS founder alleles.
The default colors for the Collaborative Cross (CC) have been changed to a color-blind friendly palette. The original CC colors remain as CCorigcolors; the previous default is now CCaltcolors. The new colors are derived from the palette in Wong (2011) Nature Methods doi:10.1038/nmeth.1618.
plot_coefCC() was revised to include col=CCcolors as an argument. The default is the new color-blind friendly CC colors, but one can now more easily use col=CCaltcolors or col=CCorigcolors to get a different choice.
Added plot_sdp() to plot the strain distribution patterns of SNPs using tracks of tick-marks for each founder strain. (Issue #163)
Added arguments sdp_panel and strain_labels to plot_snpasso() so that you can include the plot_sdp() panel with the SNP association results and/or the genes.
Added replace_ids() for a matrix or data frame (using the row names as the individual IDs). (Issue #191)
Have calc_het() give an error if the input are for allele dosages. (Issue #190)
Sneaky change in ind_ids() makes it apply to calc_genoprob and fst_genoprob objects. I’m not sure how to document this. (Issue #189)
The output of est_herit() now includes the residual SD as an attribute, "resid_sd". (Issue #16)
Implemented a cross type "hsf1" that is similar to "dof1", for a cross between an 8-way HS individual and a 9th strain. (Issue #149)
calc_kinship() died with cryptic error if genotype probabilities didn’t have a names attribute; now using seq_len(probs).
Give better error messages in est_map(), viterbi(), and sim_geno() if the cross is missing the genetic map.
Fixed Issue #194: calc_genoprob() was taking chromosome names from cross$gmap which might have been missing; now using names(cross$geno).
Fixed Issue #195: in create_snpinfo(), drop markers that are non-informative.
Fixed Issue #196, that step() returns -Inf rather than NaN for general AIL. This had to do with the handling of -Inf in addlog().
In fit1() and scan1coef(), wasn’t grabbing the ... arguments. properly.
Ugly c++ revisions to avoid clang UBsan warnings on CRAN. (Issue #169)
Revised reduce_markers() so that it can handle the case of many markers, by working with them in smaller batches.
fit1() now returns both fitted values and residuals.
fit1() can be run with genotype probabilities omitted, in which case an intercept column of 1’s is used (Issue #151).
Updated mouse gene database with 2020-09-07 data from MGI.
Implemented Issue #184, to make calc_het() multi-core.
Make the vdiffr package optional: only test the plots locally, and only if vdiffr is installed.
calc_sdp() can now take a plain vector (Issue #142).
Added a lodcolumn argument to maxlod() (Issue #137).
Fixed Issue #181, where calc_het() gave values > 1 when used with R/qtl2fst-based probabilities. Also fixed a similar bug in calc_geno_freq().
Fixed Issue #172, where fit1() gave incorrect fitted values when kinship is provided, because they weren’t “rotated back”.
fit1() no longer provides individual LOD scores (ind_lod) when kinship is used, as with the linear mixed model, the LOD score is not simply the sum of individual contributions.
Fixed Issue #174 re genoprob_to_alleleprob() for general AIL crosses. We had not implemented the geno2allele_matrix() function.
Fixed Issue #164, so plot_pxg() can handle a phenotype that is a single-column data frame.
Fixed Issue #135, so plot_scan1() can take vector input (which is then converted to a single-column matrix).
Fixed Issue #157, to have calc_genoprob() give a better error message about missing genetic map.
Fixed Issue #178, to have read_cross2() give a warning not an error if incorrect number of alleles.
Fixed Issue #180 re scan1() error if phenotypes’ rownames have rownames.
Fixed Issue #146, revising predict_snpgeno() so that it works for homozygous populations, like MAGIC lines or the Collaborative Cross.
Fixed Issue #176, that guess_phase() doesn’t work with cross type "genail". Needed to define phase_known_crosstype as "genailpk" in cross_genail.h because otherwise is_phase_known() will return TRUE.
Fixed compilation error on Solaris on CRAN, due to a log(10) and sqrt(10). Solaris refuses to do log(int) or sqrt(int), I guess.
Fixed some conflicting enum definitions in c++ code
Added recognition of the R Core Team as a contributor, as the package includes a copy of code for Brent’s method for univariate function optimization. Also added a Copyright field in the DESCRIPTION field, explaining the copyright of that code.
Sped up some of the examples and tests. Tests no longer use more than 2 cores (even those that are only run locally).
Added \value{} sections in the documentation for various functions. Added further explanation of "viterbi", "scan1", "scan1perm", etc. S3 classes in the documentation.
Added some functions for diagnostics: recode_snps(), calc_raw_het(), calc_raw_geno_freq(), calc_raw_maf(), and calc_raw_founder_maf().
Added argument blup to fit1(), for getting BLUPs for a single fixed QTL position. At present, just gives estimates and coefficients by calling scan1blup() with a single position.
pull_genoprobpos() can now take either a marker name (as before) or a set of map, chromosome, and position (from which it uses find_marker() to get the marker name).
Added plot function for the results of compare_geno(). (Plots histogram of upper triangle.)
Added functions n_founders() and founders() for getting the number of founders and the founder strain names for a cross2 object.
scan1() now takes an optional hsq argument, so that the residual heritability may be specified rather than estimated.
write_control_file() now allows cross info codes with a cross info file (previously only allowed with a covariate). read_cross2() gives a warning if there are cross info conversion codes but more than one cross info column.
Small fix in read_cross2() to allow multiple cross info covariates.
Added a check that the founder genotypes have the same strain IDs on each chromosome.
convert2cross2() now includes alleles component even if it wasn’t present as an attribute.
Added function sdp2char() for converting numeric SDP codes to character strings like "ABC|DEFGH".
Updated mouse gene database with 2019-08-12 data from MGI.
get_common_ids() strips off names from output, just in case.
Added internal functions rqtl1_crosstype() and rqtl1_chrtype().
Fixed typo in help for scan1() and related functions.
genoprob_to_snpprob() was giving an error if you gave a cross2 object in place of a snpinfo table and it had monomorphic markers.
Fixed problem with weights in scan1() and related functions when their derived from table(). Make sure they’re a plain numeric vector, not an array.
Fixed check_cross2(): the check for invalid genotypes wasn’t happening.
Better error message for the case that there are no markers in common between map and genotypes.
extract_dim_from_header(), used by read_cross2() and read_csv(), now just looks for the number part in the rest of the line.
maxlod() now handles missing values (forcing na.rm=TRUE). If all values are missing it gives a warning and returns -Inf. [Fixes Issue #134.]
In max_scan1(), treat the case that the input has no column names. [Fixes Issue #133.]
max_scan1() was giving a messed up error message if lodcolumn was out of range. [Fixes Issue #132.]
Revised the script inst/scripts/create_ccvariants.R to capture all of the consequences and genes for each SNP (rather than just the first), and fixing a bug that prevented capture of indels from chromosomes 6-X. Consequently, revised the example SQLite database extdata/cc_variants_small.sqlite and associated tests.
scan1coef() and fit1() now, by default, gives coefficient estimates for the QTL effects that sum to 0, with an additional coefficient being the intercept. This makes it more like DOQTL (and scan1blup()). The previous behavior can be obtained with the argument zerosum=FALSE.
Added function create_snpinfo() for creating a SNP information table from a cross2 object, for use with scan1snps().
Updated extdata/mouse_genes_small.sqlite using updated MGI annotations. Some of the field names have changed.
In check_cross2(), added a test for alleles being a vector of character strings.
Fixed some tests for R 3.6, due to change in random number generation.
Use Markdown for function documentation, throughout
In genoprob_to_snpprob() when a cross object is provided, make sure the genotype probabilities get subset to the cross markers.
Fixed bug in scan1snps() re keep_all_snps=FALSE. It wasn’t subsetting to the index SNPs properly. Added an internal function reduce_to_index_snps(). (See Issue #89.)
Fixed bug in step probabilities for 4-, 8-, and 16-way RIL by selfing.
Fixed bug in zip_datafiles() when the files are in a subdirectory. (See Issue #102.)
Fixed bug in plot_peaks() for the case that the input peaks object does not contain QTL intervals. (See Issue #107.)
Fixed inappropriate warning message for check of cross_info with cross type risib8.
Fixed bugs in guess_phase() and locate_xo() where we needed an any() around a comparison of two vectors.
Added plot_lodpeaks() for scatterplot of LOD score vs position for inferred QTL from find_peaks() output.
Added new cross types "genril" and "genail", implemented to handle any number of founders; include the number of founders in the cross type, for example "genril38" or "genail38". The cross information has length 1 + number of founders, with first column being the number of generations and the remaining columns being non-negative integers that indicate the relative frequencies of the founders in the initial population (these will be scaled to sum to 1). "genril" assumes the progeny are inbred lines (recombinant inbred lines, RIL), while "genail" assumes the progeny have two random chromosomes (advanced intercross lines, AIL).
The internal function batch_vec() now made user-accessible, and takes an additional argument n_cores. This splits a vector into batches for use in parallel calculations.
The internal function cbind_expand() now made user-accessible. It’s for combining matrices using row names to align the rows and expanding with missing values if there are rows in some matrices but not others.
In plot_peaks(), added lod_labels argument. If TRUE, include LOD scores as text labels in the figure.
Added function calc_het() for calculating estimated heterozygosities, by individual or by marker, from genotype probabilities derived by calc_genoprob().
Small corrections to documentation.
Revise some tests due to change in Recla and DOex datasets at https://github.com/rqtl/qtl2data
Add tests of decomposed kinship matrix (from decomp_kinship()) with scan1().
rbind_scan1() and cbind_scan1() no longer give error if inputs don’t all have matching attributes.
Change default gap between chromosomes in plot_scan1() (and related) to be 1% of the total genome length.
Fixed bug in subset_kinship() that prevented scan1() from working with decomposed “loco” kinship matrices.
Fixed descriptions in help files for cbind.calc_genoprob() and rbind.calc_genoprob(), for column- and row-binding genotype probabilities objects (as output by calc_genoprob(). cbind() is for the same set of individuals but different chromosomes. rbind() is for the same set of markers and genotypes but different individuals. Made similar corrections for the related functions for sim_geno() and viterbi() output.
Added pull_genoprobint() for pulling out the genotype probabilities for a given genomic interval. Useful, for example, to apply scan1blup() over a defined interval rather than an entire chromosome.
scan1(), scan1perm(), scan1coef(), fit1(), and scan1snps() can now use weights when kinship is provided, for example for the case of the analysis of recombinant inbred line (RIL) phenotype means with differing numbers of individuals per line. The residual variance matrix is like \(v[h^2 K + (1-h^2)D]\) where D is diagonal {1/w} for weights w.
Add weights argument to est_herit().
Added add_threshold() for adding significance thresholds to a genome scan plot.
Added predict_snpgeno() for predicting SNP genotypes in a multiparent populations, from inferred genotypes plus the founder strains’ SNP alleles.
In genoprob_to_snpprob(), the snpinfo argument can now be a cross object (for a multiparent population with founder genotypes), in which case the SNP information for all SNPs with complete founder genotype data is calculated and used.
max_scan1() with lodcolumn=NULL returns the maximum for all lod score columns. If map is included, the return value is in the form returned by find_peaks(), namely with lodindex and lodcolumn arguments added at the beginning.
Added replace_ids() for replacing individual IDs in an object. S3 method for "cross2" objects and output of calc_genoprob(), viterbi(), maxmarg(), and sim_geno().
Added clean_scan1() plus generic function clean() that works with both this and with clean_genoprob(). clean_scan1() replaces negative values with NA and removes rows that have all NAs.
More informative error message in est_herit(), scan1(), etc., when covariates and other data are not numeric.
Fixed pull_genoprobpos() so it will work with qtl2feather (and qtl2fst).
In plot_genes(), if xlim is provided as an argument, subset the genes to those that will actually appear in the plotting region.
Revise find_marker() so that the input map can also be a “snp info” table (with columns "snp_id", "chr" and "pos").
Added find_index_snp() for identifying the index SNP that corresponds to a particular SNP in a snp info table that’s been indexed with index_snps().
Add overwrite argument (default FALSE) to zip_datafiles(), similar to that for write_control_file().
plot_snpasso() now takes an argument chr.
max_scan1() no longer gives a warning if map is not provided.
insert_pseudomarkers() will now accept pseudomarker_map that includes only a portion of the chromosomes.
In fit1(), replaced tol and maxit and added ... which takes these plus a few additional hidden control parameters.
Fix a bug in index_snps(); messed up results when start and end outside the range of the map.
Fix a bug in scan1snps() regarding use of chr argument: need to force to be unique character strings, and avoid unnecessary warning about start and end.
Fix a bug in scan1snps() where it didn’t check that the genoprobs and map conform.
Revised underlying binary trait regression function to avoid some of the tendency towards NAs.
Added function clean_genoprob() which cleans genotype probabilities by setting small values to 0 and, for genotype columns where the maximum value is not large, setting all values to 0. This is intended to help with the problem of unstable estimates of genotype effects in scan1coef() and fit1() when there’s a genotype that is largely absent.
Added function compare_maps() for comparing marker order between two marker maps.
Revised the order of arguments in reduce_markers() to match pick_marker_subset(), because I like the latter better. Removed the function pick_marker_subset() because it’s identical to reduce_markers(). (Seriously, I implemented the same thing twice.)
plot_coef() now uses a named lodcolumn argument, if provided, to subset scan1_output, if that’s provided.
In the documentation for scan1coef(), scan1blup(), and fit1(), revised the suggested contrasts for getting additive and dominance effects in an intercross.
plot_coef() with scan1_output provided, ylim_lod was being ignored.find_peaks() and max_scan1() can now take snpinfo tables (as produced by index_snps() and scan1snps()) in place of the map.find_peaks().find_peaks(). (Stopped with error if no LOD scores were above the threshold.)The output of fit1() now includes fitted values.
Added function pull_genoprobpos() for pulling out a specific position (by name or position) from a set of genotype probabilities.
The chr column in the result of find_peaks() is now a factor. This makes it possible to sort by chromosome. Also added an argument sort_by for choosing how to sort the rows in the result (by column, genomic position, or LOD score).
In max_scan1(), if map is not provided, rather than stopping with an error, we just issue a warning and return the genome-wide maximum LOD score.
Revised find_markerpos() so it can take a map (as a list of vectors of marker positions) in place of a "cross2" object.
plot.scan1(), which failed to pass lodcolumn to plot_snpasso().The previously separate packages qtl2geno, qtl2scan, qtl2plot, and qtl2db have now been combined into one package.