public static Tuple <string, string, int, string>[] FromResnResi // (operation, wt_resn, wt_resi, mut_resn)[]
(Tuple <string, int>[] wt_resn_resi
, Tuple <string, int>[] mut_resn_resi
, Tuple <int, int> rngLcsLen
)
{
var lcs = HSequence.LongCommSubseq.GetLongCommSubseq(wt_resn_resi, mut_resn_resi);
if (rngLcsLen != null)
{
if ((lcs.lcs_resn.Length < rngLcsLen.Item1) || (rngLcsLen.Item2 < lcs.lcs_resn.Length))
{
return(null);
}
}
List <Tuple <string, string, int, string> > mutationinfos = new List <Tuple <string, string, int, string> >();
// (operation, wt_resn, wt_resi, mut_resn)
//var aminoacids = HBioinfo.AminoAcids.ToDictionaryBy3Letter(true);
//string[] mutate = new string[0];
for (int i = 0; i < lcs.lcs_oper1to2.Length; i++)
{
if (lcs.lcs_oper1to2[i].Item1 == LCS.Oper.Match)
{
continue;
}
/// 1. collecting deletes and inserts until next match
List <int> idxs_del = new List <int>();
List <int> idxs_ins = new List <int>();
int last_j = i;
for (int j = i; j < lcs.lcs_oper1to2.Length; j++)
{
if (lcs.lcs_oper1to2[j].Item1 == LCS.Oper.Match)
{
break;
}
if (lcs.lcs_oper1to2[j].Item1 == LCS.Oper.Delete)
{
idxs_del.Add(j);
}
if (lcs.lcs_oper1to2[j].Item1 == LCS.Oper.Insert)
{
idxs_ins.Add(j);
}
last_j = j;
}
/// 2. determine matching del-ins
List <Tuple <int, int> > idxs_del_ins = new List <Tuple <int, int> >();
while ((idxs_del.Count != 0) && (idxs_ins.Count != 0))
{
int idx_del = (i == 0) ? idxs_del.Last() : idxs_del.First();
int idx_ins = idxs_ins.First();
idxs_del_ins.Add(new Tuple <int, int>(idx_del, idx_ins));
HDebug.Verify(idxs_del.Remove(idx_del));
HDebug.Verify(idxs_ins.Remove(idx_ins));
}
/// 3. handle mutation informations
while ((idxs_del.Count != 0) || (idxs_ins.Count != 0) || (idxs_del_ins.Count != 0))
{
Tuple <int, int> idx_del_ins = null;
int idx_del = int.MaxValue; if (idxs_del.Count > 0)
{
idx_del = idxs_del.First();
}
int idx_ins = int.MaxValue; if (idxs_ins.Count > 0)
{
idx_ins = idxs_ins.First();
}
int idx_rep = int.MaxValue; if (idxs_del_ins.Count > 0)
{
idx_del_ins = idxs_del_ins.First(); idx_rep = idx_del_ins.HToArray().Min();
}
int min_idx = HMath.HMin(idx_del, idx_ins, idx_rep);
HDebug.Assert(min_idx != int.MaxValue);
if (idx_del == min_idx)
{
string resn1 = lcs.lcs_oper1to2[idx_del].Item2;
int resi1 = lcs.lcs_oper1to2[idx_del].Item3.Value;
//string mutatei = string.Format("delete {0}{1}", aminoacids[resn].name1, resi);
//mutate = mutate.HAdd(mutatei);
mutationinfos.Add(new Tuple <string, string, int, string>("delete", resn1, resi1, null));
HDebug.Verify(idxs_del.Remove(idx_del));
continue;
}
if (idx_ins == min_idx)
{
string resn_insertafter = (i == 0) ? " " : lcs.lcs_oper1to2[i - 1].Item2;
int resi_insertafter = (i == 0) ? int.MinValue : lcs.lcs_oper1to2[i - 1].Item3.Value;
string resn2 = lcs.lcs_oper1to2[idx_ins].Item2;
HDebug.Assert(lcs.lcs_oper1to2[idx_ins].Item3 == null);
mutationinfos.Add(new Tuple <string, string, int, string>("insert", resn_insertafter, resi_insertafter, resn2));
HDebug.Verify(idxs_ins.Remove(idx_ins));
continue;
}
if (idx_rep == min_idx)
{
int idel = idx_del_ins.Item1;
int iins = idx_del_ins.Item2;
string resn1 = lcs.lcs_oper1to2[idel].Item2;
int resi1 = lcs.lcs_oper1to2[idel].Item3.Value;
string resn2 = lcs.lcs_oper1to2[iins].Item2;
HDebug.Assert(lcs.lcs_oper1to2[iins].Item3 == null);
//string mutatei = string.Format("{0}{1}{2}", aminoacids[resn1].name1, resi1, aminoacids[resn2].name1);
//mutate = mutate.HAdd(mutatei);
mutationinfos.Add(new Tuple <string, string, int, string>("replace", resn1, resi1, resn2));
HDebug.Verify(idxs_del_ins.Remove(idx_del_ins));
continue;
}
}
i = last_j;
continue;
}
return(mutationinfos.ToArray());
}
