public static void GenerateMatrix()
{
var sites = CleavageSiteUtility.Deserialize();
float[,] matrix = new float[sites.Count, 21];
for (int i = 0; i < sites.Count; i++)
{
System.Console.WriteLine(sites[i].Gene);
for (int j = 0; j < 21; j++)
{
matrix[i, j] = Reactivity.GetReactivity(sites[i].Gene, sites[i].StartAt + j - 1);
}
}
//print
string content = string.Empty;
for (int i = 0; i < matrix.GetLength(0); i++)
{
for (int j = 0; j < matrix.GetLength(1); j++)
{
content += matrix[i, j].ToString() + ",";
}
content = content.TrimEnd(',') + "\n";
}
FileExtension.Save(content, Config.WorkingFolder + "cs_21_reactivity.csv");
}
private static void GenerateShapeFile(CleavageSite site, int startAt, int endAt, string fileName)
{
string content = string.Empty;
for (int position = startAt; position < endAt; position++)
{
content += $"{position - startAt + 1} {Reactivity.GetReactivity(site.Gene, position)}\r\n";
}
FileExtension.Save(content, fileName);
}
/// <summary>
/// Genes that can be found in the cleavage efficiency files and the reactivity file.
/// </summary>
public static string[] GetValidNamesAll()
{
if (File.Exists(Config.WorkingFolder + "genes_have_all.txt"))
{
return(FileExtension.ReadList(Config.WorkingFolder + "genes_have_all.txt"));
}
string[] csNames = GetValidNames();
string[] eNames = Degradome.GetValidNames();
string[] rNames = Reactivity.GetValidNames();
string[] intersection = csNames.Intersect(eNames).Intersect(rNames).ToArray();
FileExtension.SaveList(Config.WorkingFolder + "genes_have_all.txt", intersection);
return(intersection);
}
public static void GenerateAll()
{
GetValidNamesAll();
Degradome.ToCleavageEfficiency();
Reactivity.Serialise();
Gene.Serialise();
GenerateCleavegeSites();
GenerateCleavageSiteFiles();
GenerateAverageReactivity();
GenerateStructureFiles();
// step 9
GenerateRnaStructPlots();
// step 10
RnaDistancePreprocess.Process();
}
private static float[][] GetAverageAndStandardDeviation(int datasetIndex, DegradomeType dType)
{
var list = ReadCleavageSite($"{Config.WorkingFolder}\\cleavage_sites_{dType}_{datasetIndex}.csv");
float[][] matrix = new float[21][];
float[][] stats = new float[2][];
//average
stats[AVG] = new float[21];
//StandardDeviation
stats[SD] = new float[21];
for (int col = 0; col < 21; col++)
{
matrix[col] = new float[list.Count];
for (int row = 0; row < list.Count; row++)
{
try
{
var site = list[row];
matrix[col][row] = Reactivity.GetReactivity(site.Gene, site.StartAt - 1 + col);
}
catch
{
}
}
}
for (int col = 0; col < 21; col++)
{
stats[AVG][col] = matrix[col].Average();
stats[SD][col] = (float)matrix[col].StandardDeviation();
}
return(stats);
}
public static void Filter()
{
var sites = CleavageSiteUtility.Deserialize();
string deletedCleavageSite = string.Empty;
List <CleavageSite> toDelete = new List <CleavageSite>();
foreach (var site in sites)
{
if (!Reactivity.HasReactivity(site.Gene))
{
toDelete.Add(site);
deletedCleavageSite += site.ToString() + "\r\n";
}
}
foreach (var site in toDelete)
{
sites.Remove(site);
}
FileExtension.Save(deletedCleavageSite, "CleavageSite_Filtered_by_Reactivity.txt");
Serializer.Serialize("cleavage_sites.bin", sites);
//return sites;
}
public static void GenerateCleavageSiteFiles()
{
List <CleavageSite> sites = new List <CleavageSite>();
string[] lines = FileExtension.ReadList(Config.WorkingFolder + "cleavage_sites.csv");
Console.WriteLine($"Total Cleavage Sites:{lines.Length}");
foreach (string line in lines)
{
var arr = line.Split('_');
CleavageSite site = CleavageSite.Parse(line);
sites.Add(site);
}
Console.WriteLine($"After filtered by cleavage efficiency and reactivity:{sites.Count}");
//Generate 4 files filtered by 4 degradome datasets.
foreach (DegradomeType dType in EnumUtil.GetValues <DegradomeType>())
{
string site_content = string.Empty;
string efficiency_content = string.Empty;
string efficiency_log_content = string.Empty;
string reactivity_content = string.Empty;
string reactivity_25_content = string.Empty;
string reactivity_50_content = string.Empty;
//string reactivity_75_content = string.Empty;
//string reactivity_100_content = string.Empty;
foreach (CleavageSite site in sites)
{
if (site.Extendability < 50)
{
break;
}
//if the cleavage efficiency is zero,
//we can say that this is not a cleavage site.
//filter it out.
float efficiency = Efficiency.GetEfficiency(site, dType);
if (efficiency == 0)
{
continue;
}
site_content += site.ToString() + "\n";
efficiency_content += efficiency.ToString() + "\n";
efficiency_log_content += Math.Log(efficiency).ToString() + "\n";
for (int j = 0; j < 21; j++)
{
reactivity_content += Reactivity.GetReactivity(site.Gene, site.StartAt - 1 + j) + ",";
}
for (int j = 0 - 25; j < 21 + 25; j++)
{
reactivity_25_content += Reactivity.GetReactivity(site.Gene, site.StartAt - 1 + j) + ",";
}
for (int j = 0 - 50; j < 21 + 50; j++)
{
reactivity_50_content += Reactivity.GetReactivity(site.Gene, site.StartAt - 1 + j) + ",";
}
reactivity_content = reactivity_content.TrimEnd(',') + "\n";
reactivity_25_content = reactivity_25_content.TrimEnd(',') + "\n";
reactivity_50_content = reactivity_50_content.TrimEnd(',') + "\n";
}
FileExtension.Save(site_content, Config.WorkingFolder + "\\cleavage_site_" + dType + ".csv");
FileExtension.Save(efficiency_content, Config.WorkingFolder + "\\cs_efficiencies_" + dType + ".csv");
FileExtension.Save(efficiency_log_content, Config.WorkingFolder + "\\cs_efficiencies_log_" + dType + ".csv");
FileExtension.Save(reactivity_content, Config.WorkingFolder + "\\cs_reactivity_" + dType + ".csv");
FileExtension.Save(reactivity_25_content, Config.WorkingFolder + "\\cs_reactivity_" + dType + "_25.csv");
FileExtension.Save(reactivity_50_content, Config.WorkingFolder + "\\cs_reactivity_" + dType + "_50.csv");
//FileExtension.Save(reactivity_75_content, Config.WorkingFoler + "\\cs_reactivity_" + dType + "_75.csv");
//FileExtension.Save(reactivity_100_content, Config.WorkingFoler + "\\cs_reactivity_" + dType + "_100.csv");
}
}