public static Dictionary <string, protInfo> RearangeColumnOrder(Dictionary <string, protInfo> dic, string fileName)
{
Dictionary <string, protInfo> res = new Dictionary <string, protInfo>();
List <string> keys = new List <string>(dic.Keys);
StreamReader f = new StreamReader(fileName);
string line = f.ReadLine();
f.Close();
string[] aux = line.Split();
int[] index = new int[aux.Length];
for (int i = 0; i < aux.Length; i++)
{
index[i] = Convert.ToInt32(aux[i]);
}
for (int j = 0; j < keys.Count; j++)
{
protInfo xx = dic[keys[j]];
List <byte> newProfile = new List <byte>();
for (int i = 0; i < xx.profile.Count; i++)
{
newProfile.Add(xx.profile[index[i]]);
}
xx.profile = newProfile;
res.Add(keys[j], xx);
}
return(res);
}
public static Dictionary <string, protInfo> RearangeColumnOrder(Dictionary <string, protInfo> dic)
{
Dictionary <string, protInfo> res = new Dictionary <string, protInfo>();
List <double> sumColumn = new List <double>();
List <string> keys = new List <string>(dic.Keys);
for (int i = 0; i < dic[keys[0]].profile.Count; i++)
{
double sum = 0;
foreach (var item in keys)
{
if (dic[item].profile[i] > 0)
{
sum += dic[item].profile[i];
}
}
//sum++;
sumColumn.Add(sum);
}
int[] index = Enumerable.Range(0, sumColumn.Count).ToArray <int>();
Array.Sort <int>(index, (b, a) => sumColumn[a].CompareTo(sumColumn[b]));
for (int j = 0; j < keys.Count; j++)
{
protInfo xx = dic[keys[j]];
List <byte> newProfile = new List <byte>();
for (int i = 0; i < xx.profile.Count; i++)
{
newProfile.Add(xx.profile[index[i]]);
}
xx.profile = newProfile;
res.Add(keys[j], xx);
}
return(res);
}
public override Dictionary <string, protInfo> GetProfile(profileNode node, string listFile, DCDFile dcd = null)
{
Dictionary <string, protInfo> res = base.GetProfile(node, listFile, dcd);
/* StreamReader r = new StreamReader("sekw");
* string line = r.ReadLine();
* int []index = new int[399];
* int counter = 0;
* while (line != null)
* {
* index[counter] = Convert.ToInt32(line);
* line = r.ReadLine();
* counter++;
* }*/
int[] index = new int[20] {
0, 74, 57, 375, 343, 174, 33, 371, 227, 330, 9, 327, 141, 160, 256, 216, 138, 385, 51, 268
}; // for proteins
// int[] index = new int[29] { 21, 64, 29, 2, 33, 57, 77, 84, 14, 44, 81, 0, 90, 74, 30, 71, 76, 8, 48, 61,86,41,3,7,15,20,23,25,51};//for rna
List <string> keys = new List <string>(res.Keys);
foreach (var item in keys)
{
List <byte> newProfile = new List <byte>();
for (int i = 0; i < index.Length; i++)
{
newProfile.Add(res[item].profile[index[i]]);
}
protInfo xx = res[item];
xx.profile = newProfile;
res[item] = xx;
}
//res = RearangeColumnOrder(res);
// res = RearangeColumnOrder(res, "C:\\Projects\\listIndex");
res = ProfileStat.RearangeStates(res, 0.51);
return(res);
}
public override Dictionary<string, protInfo> GetProfile(profileNode node, string fileName, DCDFile dcd)
{
Dictionary<string, protInfo> dic = new Dictionary<string, protInfo>();
StreamReader wr;
DebugClass.WriteMessage("profile" + fileName);
wr = new StreamReader(fileName);
protInfo info;
string line = wr.ReadLine();
string name = "";
string seq = "";
List<string> profile = new List<string>();
List<byte> newProfile = new List<byte>();
while (line != null)
{
if (line.Contains(">"))
{
if (name.Length > 0)
{
info = new protInfo();
info.sequence = null;
info.profile = newProfile;
dic.Add(name, info);
}
name = line.Replace(">", "");
line = wr.ReadLine();
}
if (line.Contains(node.profName+" "))
{
string cLine=line.Remove(0, (node.profName+" profile ").Length);
cLine = Regex.Replace(cLine, @"\s+", " ");
cLine = cLine.Trim();
string[] aux;
if (cLine.Contains(' '))
aux = cLine.Split(' ');
else
{
char[] charArray = cLine.ToCharArray();
aux = cLine.Select(x => x.ToString()).ToArray();
}
profile.Clear();
foreach (var item in aux)
profile.Add(item);
newProfile = new List<byte>();
for (int i = 0; i < profile.Count; i++)
if (node.ContainsState(profile[i].ToString()))
newProfile.Add(node.GetCodedState(node.states[profile[i].ToString()]));
else
if(profile[i].ToString()!="-" && profile[i].ToString()!="X")
throw new Exception("Unknow state " + profile[i].ToString() + " in "+ node.profName + " profile!");
else
newProfile.Add(0);
}
line = wr.ReadLine();
}
info = new protInfo();
info.sequence = seq;
info.profile = newProfile;
dic.Add(name, info);
DebugClass.WriteMessage("number of profiles " + dic.Keys.Count);
wr.Close();
return dic;
}
public override Dictionary <string, protInfo> GetProfile(profileNode node, string fileName, DCDFile dcd)
{
Dictionary <string, protInfo> dic = new Dictionary <string, protInfo>();
StreamReader wr;
DebugClass.WriteMessage("profile" + fileName);
wr = new StreamReader(fileName);
protInfo info;
string line = wr.ReadLine();
string name = "";
string seq = "";
List <string> profile = new List <string>();
List <byte> newProfile = new List <byte>();
while (line != null)
{
if (line.Contains(">"))
{
if (name.Length > 0)
{
info = new protInfo();
info.sequence = null;
info.profile = newProfile;
if (dic.ContainsKey(name))
{
wr.Close();
throw new Exception("The nameof profile must be unique, name: " + name + " already exists in " + fileName);
}
dic.Add(name, info);
}
newProfile = new List <byte>();
name = line.Replace(">", "");
line = wr.ReadLine();
}
if (line != null && line.Contains(" profile "))
{
int index = line.IndexOf(" profile ");
if (index == -1)
{
wr.Close();
throw new Exception("Incorrect file format you have to have 'profile' word!");
}
string cLine = line.Remove(0, " profile ".Length + index);
cLine = Regex.Replace(cLine, @"\s+", " ");
if (line.Length == 0)
{
continue;
}
cLine = cLine.Trim();
string[] aux;
if (cLine.Contains(' '))
{
aux = cLine.Split(' ');
}
else
{
char[] charArray = cLine.ToCharArray();
aux = cLine.Select(x => x.ToString()).ToArray();
}
profile.Clear();
foreach (var item in aux)
{
profile.Add(item);
}
newProfile = new List <byte>();
for (int i = 0; i < profile.Count; i++)
{
if (node.ContainsState(profile[i].ToString()))
{
newProfile.Add(node.GetCodedState(node.states[profile[i].ToString()]));
}
else
if (profile[i].ToString() != "-" && profile[i].ToString() != "X")
{
throw new Exception("Unknow state " + profile[i].ToString() + " in " + node.profName + " profile!");
}
else
{
newProfile.Add(0);
}
}
}
line = wr.ReadLine();
}
if (newProfile.Count > 0)
{
info = new protInfo();
info.sequence = seq;
info.profile = newProfile;
dic.Add(name, info);
}
DebugClass.WriteMessage("number of profiles " + dic.Keys.Count);
wr.Close();
return(dic);
}
public override Dictionary <string, protInfo> GetProfile(profileNode node, string fileName, DCDFile dcd)
{
Dictionary <string, protInfo> dic = new Dictionary <string, protInfo>();
StreamReader wr;
DebugClass.WriteMessage("profile" + fileName);
wr = new StreamReader(fileName);
protInfo info;
string line = wr.ReadLine();
string name = "";
string seq = "";
List <string> profile = new List <string>();
List <byte> newProfile = new List <byte>();
while (line != null)
{
if (line.Contains(">"))
{
if (name.Length > 0)
{
info = new protInfo();
info.sequence = null;
info.profile = newProfile;
dic.Add(name, info);
}
name = line.Replace(">", "");
line = wr.ReadLine();
}
if (line.Contains(" profile "))
//if (line.Contains(node.profName+" "))
{
string[] aux;
aux = line.Split(' ');
node.profName = aux[0];
string cLine = line.Remove(0, (node.profName + " profile ").Length);
cLine = Regex.Replace(cLine, @"\s+", " ");
cLine = cLine.Trim();
if (cLine.Contains(' '))
{
aux = cLine.Split(' ');
}
else
{
char[] charArray = cLine.ToCharArray();
aux = cLine.Select(x => x.ToString()).ToArray();
}
profile.Clear();
foreach (var item in aux)
{
profile.Add(item);
}
newProfile = new List <byte>();
for (int i = 0; i < profile.Count; i++)
{
if (node.ContainsState(profile[i].ToString()))
{
newProfile.Add(node.GetCodedState(node.states[profile[i].ToString()]));
}
else
if (profile[i].ToString() != "-" && profile[i].ToString() != "X")
{
throw new Exception("Unknow state " + profile[i].ToString() + " in " + node.profName + " profile!");
}
else
{
newProfile.Add(0);
}
}
}
line = wr.ReadLine();
}
info = new protInfo();
info.sequence = seq;
info.profile = newProfile;
dic.Add(name, info);
DebugClass.WriteMessage("number of profiles " + dic.Keys.Count);
wr.Close();
return(dic);
}
public override Dictionary <string, protInfo> GetProfile(profileNode node, string fileName, DCDFile dcd = null)
{
Dictionary <string, protInfo> dic = new Dictionary <string, protInfo>();
currentProgress = 0;
if (!File.Exists(GetProfileFileName(fileName)))
{
throw new Exception("Profile rnaSS does not exists!");
}
string profileName;
if (node.ContainsState("+"))
{
profileName = LWprofile;
}
else
{
profileName = SSprofile;
}
StreamReader wr = new StreamReader(GetProfileFileName(fileName));
List <string> files = GetFileList(fileName);
maxV = files.Count;
Dictionary <string, int> dicFile = new Dictionary <string, int>();
foreach (var item in files)
{
string[] aux = item.Split(Path.DirectorySeparatorChar);
if (!dicFile.ContainsKey(aux[aux.Length - 1]))
{
dicFile.Add(aux[aux.Length - 1], 0);
}
}
protInfo info;
string line = wr.ReadLine();
string name = "";
string seq = "";
List <string> profile = new List <string>();
List <byte> newProfile = new List <byte>();
while (line != null)
{
if (line.Contains('>'))
{
if (name.Length > 0)
{
info = new protInfo();
info.sequence = seq;
info.profile = newProfile;
if (dicFile.ContainsKey(name))
{
if (!dic.ContainsKey(name))
{
dic.Add(name, info);
}
else
{
ErrorBase.AddErrors("There are two the same names " + name + " in the profile file " + GetProfileFileName(fileName));
}
}
currentProgress++;
}
name = line.Remove(0, 1);
}
if (line.Contains(SEQprofile))
{
seq = line.Remove(0, SEQprofile.Length);
}
if (line.Contains(profileName))
{
string tmp = line.Remove(0, profileName.Length);
string[] aux;
profile.Clear();
if (tmp.Contains(" "))
{
tmp = tmp.TrimEnd();
aux = tmp.Split(' ');
foreach (var item in aux)
{
profile.Add(item);
}
}
else
{
for (int i = 0; i < tmp.Length; i++)
{
profile.Add(tmp[i].ToString());
}
}
newProfile = new List <byte>();
for (int i = 0; i < profile.Count; i++)
{
if (node.ContainsState(profile[i].ToString()))
{
newProfile.Add(node.GetCodedState(node.states[profile[i].ToString()]));
}
else
if (profile[i].ToString() != "-" && profile[i].ToString() != "X")
{
throw new Exception("Unknow state: " + profile[i].ToString() + " in structure " + name + " in profile " + node.profName + " profile!");
}
else
{
newProfile.Add(0);
}
}
}
line = wr.ReadLine();
}
info = new protInfo();
info.sequence = seq;
info.profile = newProfile;
if (!dic.ContainsKey(name))
{
dic.Add(name, info);
}
else
{
ErrorBase.AddErrors("There are two the same names " + name + " in the profile file " + GetProfileFileName(fileName));
}
wr.Close();
currentProgress = maxV;
return(dic);
}
private void AlignProfile(string protName,string profileName,List<byte>prof)
{
int m = 0;
string alignProfile = align[protName];
List<byte> ll = new List<byte>(alignProfile.Length);
for (int i = 0; i < alignProfile.Length; i++)
{
if (alignProfile[i] == '-')
{
ll.Add(0);
continue;
}
if (m < prof.Count)
ll.Add(prof[m++]);
else
ErrorBase.AddErrors("Profile " + profileName + " for " + protName + " seems to be incorect");
}
protInfo tmp = new protInfo();
if (r.profiles[profileName].ContainsKey(protName))
{
tmp = r.profiles[profileName][protName];
tmp.alignment = ll;
r.profiles[profileName][protName] = tmp;
}
}
public void AddStructureToAlignment(string protName,string profileName,ref protInfo profile)
{
MAlignment al = new MAlignment(refSeq.Length);
string resSeq = (al.Align(refSeq, profile.sequence)).seq2;
if (!align.ContainsKey(protName))
align.Add(protName, resSeq);
if(!r.profiles[profileName].ContainsKey(protName))
r.profiles[profileName].Add(protName,profile);
AlignProfile(protName, profileName, profile.profile);
profile.alignment = r.profiles[profileName][protName].alignment;
}
public override Dictionary <string, protInfo> GetProfile(profileNode node, string listFile, DCDFile dcd = null)
{
Dictionary <string, protInfo> dic = null;
StreamReader wr;
DebugClass.WriteMessage("start reading file" + listFile);
string useFile = listFile;
List <string> names = GetFileList(listFile);
if (names == null)
{
DebugClass.WriteMessage("After GetFileList null");
}
else
{
DebugClass.WriteMessage("After GetFileList " + names.Count);
}
maxV = names.Count;
Dictionary <string, int> dicNames = new Dictionary <string, int>(names.Count);
foreach (var item in names)
{
string[] aux = item.Split(Path.DirectorySeparatorChar);
if (!dicNames.ContainsKey(aux[aux.Length - 1]))
{
dicNames.Add(aux[aux.Length - 1], 0);
}
}
DebugClass.WriteMessage("After loop");
string[] dotFlag = names[0].Split('|');
bool flag = false;
if (dotFlag.Length == 2)
{
flag = true;
}
DebugClass.WriteMessage("Before profile");
if (dcd == null)
{
wr = new StreamReader(GetProfileFileName(listFile));
}
else
{
wr = new StreamReader(GetProfileFileName(dcd));
}
DebugClass.WriteMessage("start ddd reading");
protInfo info;
string line = wr.ReadLine();
string name = "";
string seq = "";
List <string> profile;;
List <byte> newProfile = null;
string profileName;
if (node.ContainsState("H"))
{
profileName = ssProfile;
}
else
{
profileName = contactProfile;
}
//Check number of elements in file
int lineLength = 0;
while (line != null)
{
if (lineLength < line.Length)
{
lineLength = line.Length;
}
line = wr.ReadLine();
}
wr.BaseStream.Position = 0;
wr.DiscardBufferedData();
profile = new List <string>(lineLength);
dic = new Dictionary <string, protInfo>(names.Count);
line = wr.ReadLine();
string remName = "";
DebugClass.WriteMessage("Starrrrr");
while (line != null)
{
if (line[0] == '>')
{
if (name.Length > 0)
{
if (dicNames.ContainsKey(name))
{
info = new protInfo();
info.sequence = seq;
info.profile = newProfile;
//string newName = line;
//newName = newName.Remove(0, 1);
if (!dic.ContainsKey(remName))
{
dic.Add(remName, info);
}
else
{
ErrorBase.AddErrors("!!In the generated profile file structure " + remName + " exists more then once!!!\nOnly the first profile will be considered!");
}
currentProgress++;
}
}
name = line.Remove(0, 1);
remName = name;
if (name.Contains("|") && !flag)
{
string[] tmp = name.Split('|');
name = tmp[0];
}
}
if (dicNames.ContainsKey(name))
{
if (line.Contains(SEQprofile))
{
seq = line.Remove(0, SEQprofile.Length);
}
else
if (line.Contains(profileName))
{
//StringBuilder tmp = new StringBuilder(line);
line = line.Remove(0, profileName.Length);
//tmp = tmp.Remove(0, profileName.Length);
//StringBuilder tmp = new StringBuilder (line.Remove(0, profileName.Length));
string[] aux;
if (line.Contains(' '))
{
aux = line.Split(' ');
}
else
{
aux = new string[line.Length];
for (int i = 0; i < line.Length; i++)
{
aux[i] = line[i].ToString();
}
}
newProfile = CreateNewProfile(node, aux);
}
}
line = wr.ReadLine();
}
if (dicNames.ContainsKey(name))
{
info = new protInfo();
info.sequence = seq;
info.profile = newProfile;
dic.Add(remName, info);
}
wr.Close();
DebugClass.WriteMessage("Reading finished");
return(dic);
}
public override Dictionary<string, protInfo> GetProfile(profileNode node, string listFile,DCDFile dcd=null)
{
Dictionary<string, protInfo> dic = null;
StreamReader wr;
DebugClass.WriteMessage("start reading file" + listFile);
string useFile = listFile;
List<string> names = GetFileList(listFile);
if(names==null)
DebugClass.WriteMessage("After GetFileList null");
else
DebugClass.WriteMessage("After GetFileList "+names.Count);
maxV = names.Count;
Dictionary<string, int> dicNames = new Dictionary<string, int>(names.Count);
foreach (var item in names)
{
string[] aux = item.Split(Path.DirectorySeparatorChar);
if(!dicNames.ContainsKey(aux[aux.Length-1]))
dicNames.Add(aux[aux.Length-1], 0);
}
DebugClass.WriteMessage("After loop");
string[] dotFlag = names[0].Split('|');
bool flag = false;
if (dotFlag.Length == 2)
flag = true;
DebugClass.WriteMessage("Before profile");
if (dcd == null)
{
wr = new StreamReader(GetProfileFileName(listFile));
}
else
wr = new StreamReader(GetProfileFileName(dcd));
DebugClass.WriteMessage("start ddd reading");
protInfo info;
string line = wr.ReadLine();
string name="";
string seq="";
List<string> profile; ;
List<byte> newProfile = null;
string profileName;
if (node.ContainsState("H"))
profileName = ssProfile;
else
profileName = contactProfile;
//Check number of elements in file
int lineLength=0;
while (line != null)
{
if (lineLength < line.Length)
lineLength = line.Length;
line=wr.ReadLine();
}
wr.BaseStream.Position = 0;
wr.DiscardBufferedData();
profile = new List<string>(lineLength);
dic = new Dictionary<string, protInfo>(names.Count);
line=wr.ReadLine();
string remName="";
DebugClass.WriteMessage("Starrrrr");
while (line != null)
{
if (line[0]=='>')
{
if (name.Length > 0)
{
if (dicNames.ContainsKey(name))
{
info = new protInfo();
info.sequence = seq;
info.profile = newProfile;
//string newName = line;
//newName = newName.Remove(0, 1);
if (!dic.ContainsKey(remName))
dic.Add(remName, info);
else
ErrorBase.AddErrors("!!In the generated profile file structure " + remName + " exists more then once!!!\nOnly the first profile will be considered!");
currentProgress++;
}
}
name = line.Remove(0, 1);
remName = name;
if(name.Contains("|") && !flag)
{
string[] tmp = name.Split('|');
name = tmp[0];
}
}
if (dicNames.ContainsKey(name))
{
if (line.Contains(SEQprofile))
seq = line.Remove(0, SEQprofile.Length);
else
if (line.Contains(profileName))
{
//StringBuilder tmp = new StringBuilder(line);
line=line.Remove(0,profileName.Length);
//tmp = tmp.Remove(0, profileName.Length);
//StringBuilder tmp = new StringBuilder (line.Remove(0, profileName.Length));
string[] aux;
if(line.Contains(' '))
aux = line.Split(' ');
else
{
aux = new string[line.Length];
for (int i = 0; i < line.Length; i++)
aux[i] = line[i].ToString();
}
newProfile = CreateNewProfile(node, aux);
}
}
line = wr.ReadLine();
}
if (dicNames.ContainsKey(name))
{
info = new protInfo();
info.sequence = seq;
info.profile = newProfile;
dic.Add(remName, info);
}
wr.Close();
DebugClass.WriteMessage("Reading finished");
return dic;
}
