The predict executables (predictb and predictb) each fill in missing data in haplotype sequences. They also can attempt to correct for recent mutations when used with a prior recent mutation probability. The executables are called as follows: predict[b|g] [-i ] [-s ] -i : input sequence file name (default: stdin) -s : input motif file name (default: stdin) -m : prob of mutation (default: 0) -c : specifies correction of known sites inferred to be wrong Both take the same haplotype input, haplotype files described in README.files. Missing values are denoted by '?', 'X', or 'x' in these files. They both also take motif files, also described in README.files, as input. As output, they produce files of the same haplotypes with values filled in or corrected based on the motif input. The difference between the programs is in the algorithms they use to predict missing sites. predictb uses a local algorithm, in which each base is predicted based on the most probable motif covering it. predictg uses a global algorithm in which it finds the most probable complete parse of each haploid sequence and then fills in all missing bases based on the one parse.