What's new in this version?

• 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.
  

• 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 prepared for upload to the CRAN repository
  
  
• Only minor fixes in documentation etc.
  

• 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.
  

• Minor cosmetic changes in printout and in graph appearance.
  
  
• Got rid of (most of?) the annoying graphics warnings that appeared when plotting result
  

• 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).
  

• 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 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