lobal homology alignment normally has a lower sensitivity, but

specificity [Pollard, et al., 2004]. A local homology alignment

an have a higher sensitivity, but a lower specificity [Pollard, et

]. The specificity means that two sequences of different families

cted to be correctly detected through an alignment. The sensitivity

at two sequences of the same family are predicted to be correctly

through an alignment.

he alignment statistic

th issue of sequence homology alignment is the alignment

There are two statistics for use so far and they are the edit distance

volutionary distance [Rizzato, et al., 2017; Marcais, et al., 2019].

distance uses a binary value to describe the distance or the

y between two residues of every aligned pair from two sequences.

utionary distance uses a mutation probability to describe the

y between two residues of every aligned pair from two sequences.

on probability quantifies the likelihood of the mutation from one

o the other through an evolution process.

are also two kinds of gap penalties, the fixed gap penalty and the

gap penalty. Using the fixed gap penalty, no matter how long an

gap segment is, each gap of the segment has the same penalty.

, using the variable gap penalty, the penalty applied will depend

ngth of a continuous segment of gaps.

e Sellers algorithm

rs algorithm is one of the earliest sequence homology alignment

ms [Sellers, 1974]. It was developed based on the dynamic

ming algorithm. The algorithm employs a dynamic programming

calculating the edit distance between a pair of aligned sequences.