What's new in this version of Haplin
Haplin version 7.3.2, uploaded 19 August, 2024
- Bug fix: Fixed a problem in data handling that could appear in dyads, if the individual id in the ped file was not unique.
- Miscellaneous updates/improvements.
- Miscellaneous updates/improvements.
- Miscellaneous improvements.
- New helper functions by Julia Romanowska, to ease data handling: showSNPnames, getChildren, getFathers, getMothers, getFullTriads, and getDyads.
- Miscellaneous improvements.
- New helper functions by Julia Romanowska, to ease data handling: showPheno, showGen, nsnps, nindiv, and nfam.
- New function by Miriam Gjerdevik, hapRelEff, with its own vignette, to compute relative efficiencies (sample sizes) for various designs.
- Miscellaneous minor fixes.
- Unfortunately, the very popular GenABEL package was taken down from CRAN on 24 May 2018 because it’s not being maintained any longer. Haplin relied for a long time on GenABEL to store and handle GWAS data. Thanks mostly to the programming of Julia Romanowska, Haplin Version 7.1.0 now adapts its data storage to the ff package. There are thus many updates to the data handling in Haplin, in particular the new functions genDataRead and genDataPreprocess. For more details, see the documentation.
- There is a bug in 6.2.0 that in some instances may cause strata to be given incorrect names in haptables from haplinSlide with the strata argument specified. Version 6.2.1 fixes that.
- The function hapPowerAsymp has been added (thanks mostly to Miriam Gjerdevik). hapPowerAsymp provides fast power calculations for a wide range of scenarios, without needing simulations.
- Fixed a small bug in hapRun from version 6.0
- Gene-environment interactions implemented in the function gxe.
- hapPower does power simulations for a range of designs, haplotype combinations, as well as gxe analyses (thanks to Miriam Gjerdevik).
Haplin version 5.5, uploaded April 24, 2015
- P-values for individual effects are now computed with more decimals
to facilitate sorting on p-values
- Two new functions snpPower and snpSamplesize,
which compute power and sample size for single SNP analyses.
- Two new functions cbindFiles and rbindFiles.
cbindFiles takes a number of files and combines them column-wise
(side-by-side), i.e. reads each file line by line, pastes corresponding
lines, then writes to outfile. rbindFiles takes a number of files and
combines them by rows, without reading the full files into memory. The
purpose is to facilitate easy merging of very large text files that may
be too big to read into R memory.
- Plus a number of minor fixes and cosmetic changes.
Haplin version 5.3, uploaded May 22, 2013
- haplin and haplinSlide will now use actual snp names taken from the
map file rather than just number them in sequence.
- Plus a number of minor fixes and cosmetic changes.
Haplin version 5.2, uploaded March 18, 2013
- Two new functions, lineByLine and convertPed, have
been added. The purpose is to facilitate easy conversion/recoding of
large text files that may be too big to read into R memory. lineByLine
can modify general text files, typically recode values, select lines,
columns, etc., whereas convertPed is designed to handle ped files in
particular.
- Minor fixes and some cosmetic changes.
- Fixed a bug (only present in version 5.0) that caused haplin to crash
if using GWAS data through GenABEL with use.missing = T.
- In haplinSlide, the default of the argument table.output is now set to
TRUE, to reduce the risk of running out of memory when running large
analyses.
- In haplinSlide, the argument slaveOutfile can now be used also when
cpus = 1.
- Haplin now allows parent-of-origin effects through the argument ‘poo = T’. The estimated fetal effects will then be split into the effects of haplotypes inherited from the mother and from the father. In addition, a ratio of the two relative risks thus obtained is also printed, with a corresponding p-value testing whether the two effects differ. Parent-of-origin effects can be combined with estimating maternal haplotype effects through ‘maternal = T’. Parent-of-origin effects can also be computed on the X-chromosome.
- Previous versions of Haplin did not allow X-chromosome effects to be
estimated in the pure case-control design. This is now implemented.
- Plus a number of minor fixes and tidy-ups.
- The function haplinSlide now provides feedback as it runs. A new
argument, slaveOutfile, can be used to redirect the feedback to a file.
- A minor bug in the GWAS data conversion caused the conversion to crash
for some types of family structures. This has now been fixed.
- The main new feature is that Haplin now provides full support for GWAS
data (through the GenABEL library data structure). Parallel processing
is integrated in the sliding window estimation and is activated by
simply choosing the number of cpus available (implemented using the
snowfall library).
- A new argument comb.sex to be used with X-chromosome analyses. It
allows analyses of males and females separately, and different ways of
combining males and females in a joint analysis, corresponding to
dose-response models or X-inactivation.
- A simple GUI for generating syntax available at haplin.fhi.no.
NOTE: implements most but not all current Haplin features.
- Haplin now accepts data from from the X chromosome, using the
arguments ‘xchrom = T’ (and ‘sex’ to specify column for sex variable).
- Using the ‘data.out’ argument, haplin can return a processed version
of the original data file, with all haplotypes estimated/imputed for all
families.
- A new function ‘haplinSlide’ runs haplin on a consecutive series of
overlapping windows, useful to scan genes with many SNPs.
- A new function ‘haptable’ creates a useful summary table of all the
important estimation results from a Haplin run.
- A new function ‘haplinTDT’ that performs various TDT tests for
comparison. Written by Øivind Skare.
- A new function ‘output’ which saves tables and figures from a Haplin
run to disk.
Haplin version 3.0.2, uploaded Jan 11, 2010
- Changes in R release 2.10 relative to 2.9 caused Haplin to crash when
n.vars > 0. This has now been fixed.
- Some minor fixes in documentation etc.
Haplin version 3.0.1, uploaded May 27, 2009
- Haplin prepared for upload to the CRAN repository
- Only minor fixes in documentation etc.
Haplin version 3.0, uploaded April 29, 2009
This is an extensive update, some of the most important changes are:
- Implemented as a proper R library. S-Plus version terminated.
- Haplin now handles the combined case-parent triad + control-parent
triad design, and also the standard case-control design.
- Updated output
- Can choose a multiplicative dose-response model if estimation of
separate single- and double dose effects is not required
- Killed a number of minor bugs.
Haplin version 2.1.1, uploaded June 12, 2006
- Minor cosmetic changes in printout and in graph appearance.
- Got rid of (most of?) the annoying graphics warnings that appeared
when plotting result
Haplin version 2.1, uploaded May 29, 2006
- Haplin now prints tests of Hardy-Weinberg equilibrium for each marker
separately.
- When using a reference category (not the default reciprocal reference)
Haplin 2.0 forgot to print the double dose estimate for the reference
category, even though it was computed and plotted in the plot. This has
now been fixed. (Thanks to S. Hussain for pointing this out).
Haplin version 2.0, uploaded Jan 13, 2006
- A change in the format function in the very latest R release (2.2.0,
Oct 06, 2005) caused the R version of Haplin 1.0 to exit prematurely.
This has been fixed in Haplin 2.0.
- This version (finally!) has an option to run with triads containing
missing data through the EM algorithm, by setting use.missing = T.
- A computation of the standard errors and individual p-values that
corrects for missing data and ambiguous haplotypes. The consequence is
that the jackknife resampling used in version 1.0 is no longer
necessary, and only one run of Haplin is needed. Thus, much less
patience is required of the user….
- The initial data handling and sorting is considerably improved, with
the result that most computations should be faster, and Haplin can now
handle more SNPs in a run. The exact number of SNPs possible to run
depends a good deal on the data and population structure, but 6-7 SNPs
should usually run fine.
Haplin version 1.0, uploaded June 1, 2005
- Haplin now uses “reciprocal” as the default reference category. This
is slightly different from the previous “population” (which still can be
used by setting reference = “population”). See the document on
parametrization and reference for more details
- A new “markers” argument that allows the user to select the markers to
be used in the analysis. The user no longer has to produce separate data
files for each marker selection. See the document on Haplin data format
for more details
- Much improved handling of rare haplotypes. In the first (non-numbered)
version this selection was rather rough. In version 1.0 the haplotype
frequencies are estimated fairly precisely before removing the rare
haplotypes, leading to less loss of data and more precise estimates. The
argument “threshold” that determines the frequency limit for retaining
haplotypes is now followed much more closely.
- Haplin now prints p-values for all specific effects, and an overall
p-value for the locus
- Haplin now reports the amount of data removed due to missing
genotypes and which families that are removed due to Mendelian
inconsistencies
- Jackknife resampling procedure (by setting resampling = “jackknife”)
adjusts the standard error to compensate for the missing information due
to unknown haplotypes