Suppose I have the following sequences with SNPs in 4th and 9th columns:

IND1 GATATTCGGA-
IND2 GATGTTCTGAA
IND3 GATATTCGGA-
IND4 GATATTCGGAA
IND5 GATGTTCTGAA
IND6 GATGTTCTGAA

Haplotypes from such sequences would be:

AG- (2/6)
GTA (3/6)
AGA (1/6)

So far so good.

Now commonly I have a PED file which only contains SNP's. Using my example above columns with only SNPs would be:

AC-
GTA
AG-
AGA
GTA
GTA

(actually the last column is not a SNP column but I hope is irrelevant because I have only SNPs as input)

So haplotyping is just a matter of counting and grouping equal sequences?

Where is the "haplotype inference" really?