When you are comparing sequences it is common to first align them using a dynamic programing algorithm. SEQPDIST uses NWALIGN to pair-wise align all possible pairs of sequences and then takes the measure from the alignment.
Consider:
seqpdist({'AACGT','AAGT','AAT'},'alpha','nt','square',1,'method','p-dist')
The alignment between 1 and 2 is 'AACGT' and 'AA-GT' =>1/5
The alignment between 1 and 3 is 'AACGT' and 'AA--T' =>2/5
The alignment between 1 and 3 is 'AAGT' and 'AA-T' =>1/4
HTH