First step

Inputs: X, window sizes wd, wm, and wf, threshold fTHRESH, sequence data S Outputs: SUSPECT, the set of candidate indel locations

For k=w f + 1, …,T − wf, where T is the length of the reference sequence, compute the

function f(k) as defined above.

Let SUSPECT ←{k : f (k) > fTHRESH}

Second step

Inputs: X, SUSPECT, S0, M

Outputs: INDEL, the set of indel locations

INDEL ← EMPTY

For all k ∈ SUSPECT do testseq ← $X_{k-40}^{k+40}$

For all yi ∈ S do

Align yi and testseq using Smith-Waterman (SW) algorithm.

end for

if more than M yi align to testseq with less than 2 mismatches and with shared indel interval [p, q], then

INDEL ← INDEL ∪ {[p, q]}

end if end for