BUGS, PITFALLS, AND FUTURE PLANS FOR IBDREG


11/2006

plot functions
----------------------------------------
1) title, each plot function looks for 'title=' in the ...
   If your title has only text, that is fine.  
   If you have your title stored in a variable, there is only one 
   way to complete this.  You have to build the call to plot.*  
   with text and paste it with the variable name within it.

2) The broken lines appear first in R, and the solid line (linkage) is
   plotted later.  Maybe fix by plotting only the region, then all the
   lines by lines().

3) The legend seems to cover up lines that reach into the legend area
   (upper left)

4) function to plot results for multiple ibdreg objects on a 4x6 page, as done for
   figures in paper, or just give script

5) different colors appear for R and S-PLUS when plotting an ibdreg
   object, and also known differences between motif and pdf.  
   Incorporate some kind of color map to regulate these colors.


ibdreg
-----------------------------------------
1) formula only processes pairSum() + pairDiff(), does not properly handle
   interactions: pairSum()*pairDiff().  It assumes all terms have a
   '+' between covariates.

2) function to handle runnign ibdreg on multiple chromosomes
   the difficulty is that there are so many large ibd files. We have
   scripts that do this, and it is a lot of system calls. Plotting all
   these objects is bullet (4) above

3) On rare occasions, the ibd.var matrices may not invert well.  We
   incorporated an epsilon parameter that controls the minimum svd values
   to use for generalized inverse.


Miscellaneous pitfalls for users
-----------------------------------------
1) use Merlin, which does not handle large pedigrees

2) find the perl scripts in /ibdreg/perl/, and make appropriate edits
   to gws.merlin.ibd.pl

4) defining your own pairOp() function, be sure it works for vectors
e.g. pairMean(x, y) { mean(x, y) }   DOES NOT WORK!!