The following haplotypes from markers at the CTLA4 locus (Johnson et al. 2001) can be read into a SAS data set as follows:

   data ctla4; 
      input (m1-m12)($) freq;
      datalines;
   C T A A G C C A C C A G 0.333
   T T A G G C C G C T G G 0.224
   T C A G G C C G C T G G 0.058
   T T A A G C C G C T G G 0.020
   C T A A G T C A C C A G 0.080
   C T A G G T C A C C A G 0.017
   C T A G G C C A C C A G 0.045
   T T A G G C C A C C A G 0.018
   C T G G A C T A T C G A 0.086
   C T G G A C C A T C G A 0.054
   C T G G A C C A C C G A 0.021
   ;

You can now use PROC HTSNP to search a subset of markers that explains most of the haplotype diversity in this sample. The following statements perform the search:

   proc htsnp data=ctla4 size=5 method=im  
              cutoff=0.05 seed=244 conv=0.99;
      var m1-m12;
      freq freq;
   run; 
   

The iterative maximization algorithm is selected as the search method with the METHOD=IM option. The SIZE=5 option indicates that only subsets containing exactly five SNPs are considered in the search. All haplotypes in the data set with a frequency below 0.05 are excluded from the search process because the CUTOFF=0.05 option was specified. The search continues until the convergence criterion of 0.99 is met as specified in the CONV= option. The iterative maximization algorithm randomly selects an initial set of markers, so using different seeds can produce different results.

The results from the procedure are shown in Figures 9.1 and 9.2.