Table 2. Referential factorization with RME prediction.

	Input: Sequence database S = {s 1,…, s n} and collection of references REF = {ref 1,…, ref m}
	Output: factorized sequence database fsd = {f 1,…, f n}
1:	Let fsd = ∅
2:	for 1 ≤ i ≤ n do
3:	Let f be an empty list
4:	Let s = s i
5:	while ∣s∣ ≠ 0 do
6:	Let candidate = predict(s, prme, REF)
7:	if candidate ≠ (0,0,0,0) then
8:	Add candidate to the end of f
9:	Remove the first ∣candidate∣ symbols from s
10:	else
11:	Let pre be the longest prefix of s, such that $(pos,pre)\in search{(re{f}_i)}_{pre}^0$, for a number pos, and there exists no 1 ≤ j ≤ n, with j ≠ i and ref j contains a longer prefix of s than ref i
12:	if s ≠ pre then
13:	Set rme = (ref i, pos,∣pre∣, s(∣pre∣))
14:	Add rme to the end of f
15:	Remove the first ∣pre∣+1 symbols from s
16:	else
17:	Set rme = (ref i, pos,∣pre∣ − 1, s(∣pre∣ − 1))
18:	Add rme to the end of f
19:	Remove the prefix pre from s
20:	end if
21:	end if
22:	Let prme = rme
23:	end while
24:	Add f to the end of fsd
25:	end for