// bin\bgzip.exe bam_chrY.vcf
// bin\tabix.exe bam_chrY.vcf.gz
// bin\bcftools.exe query -f '%POS\t[%TGT]\n' -r chrY:3131128-3131246 bam_chrY.vcf.gz > DYS393.txt
static void Main(string[] args)
{
StringBuilder report = new StringBuilder();
if (args.Length != 3)
{
Console.WriteLine("*** This program is not designed to work independently. Use it at your own risk ***");
Console.WriteLine("ystrfinder.exe <ref> <in_vcf_bgz> <out_report_html>");
return;
}
string yref_file = args[0]; //"ystr_info.txt";
string intput_file = args[1]; // @"C:\Genetics\Y-STR Kit\bam_chrY.vcf.gz";
string out_file = args[2];
if (!File.Exists(yref_file) || !File.Exists(intput_file))
{
Console.WriteLine("input/ref does not exist!");
return;
}
YSTRLib ylib = new YSTRLib(intput_file);
string[] lines = File.ReadAllLines(yref_file);
string[] data = null;
string ystr = null;
long begin_pos = 0;
long end_pos = 0;
string regex_motif = null;
string filter_motif = null;
string replace_regex = null;
String rpt_table_overview = Y_STR_Finder.Properties.Resources.ystr_overview;
report.Append("<html><head><title>Y-STR Kit Report</title></head><body style='font-family:Calibri; Color:black; font-style:normal'><h1>Y-STR Kit Report</h1>");
report.Append("@TABLE@");
foreach (string line in lines)
{
if (line.StartsWith("#") || line.Trim() == "")
{
continue;
}
data = line.Split(new char[] { '\t' });
if (data[0].StartsWith("TITLE"))
{
report.Append("<h2>" + data[1] + "</h2>");
continue;
}
ystr = data[0];
begin_pos = long.Parse(data[1]);
end_pos = long.Parse(data[2]);
regex_motif = data[3];
filter_motif = data[4];
replace_regex = data[5];
YStrResult result = ylib.getYSTR(begin_pos, end_pos, regex_motif, filter_motif, replace_regex);
//Console.WriteLine(ystr + " = " + result.value+"\t["+result.reliability+"]");
report.Append("<a name='" + ystr + "'></a>");
report.Append("<h3>" + ystr + " = " + result.value + "</h3>");
report.Append("<i>Location: ChrY:" + begin_pos + "-" + end_pos + "</i><br>");
report.Append("<div style='width:600px; padding:10px;word-wrap: break-word;color:#909090;font-family: monospace;'>" + result.html_seq + "</div>");
report.Append("<i>Reliability is " + result.reliability + "%</i><br><br>");
rpt_table_overview = rpt_table_overview.Replace("@" + ystr + "@", result.value.ToString());
if (result.reliability == 100)
{
rpt_table_overview = rpt_table_overview.Replace("@" + ystr + "_COLOR@", "green");
}
else
{
rpt_table_overview = rpt_table_overview.Replace("@" + ystr + "_COLOR@", "red");
}
//Console.WriteLine(result.raw_seq);
//Console.WriteLine();
}
report.Append("<i>Report generated by <a href='http://www.y-str.org/'>Y-STR Kit</a>.</i>");
report.Append("</body></html>");
string final_report = report.ToString();
final_report = final_report.Replace("@TABLE@", rpt_table_overview);
File.WriteAllText(out_file, final_report);
}
private YStrResult getYSTR(string bcfout_file, string motif_regex, string filter_regex, string replace_regex, long pos_start, long pos_end)
{
YStrResult result = new YStrResult();
try
{
string[] data = null;
long pos = 0;
string allele = "";
Dictionary <long, string> sequence = new Dictionary <long, string>();
string tmp = "";
foreach (string line in File.ReadLines(bcfout_file))
{
data = line.Split(new char[] { '\t' });
pos = long.Parse(data[0]);
allele = data[1].Split(new char[] { '/' })[0].Trim();
if (sequence.ContainsKey(pos))
{
tmp = sequence[pos];
if (tmp.Length < allele.Length)
{
sequence.Remove(pos);
sequence.Add(pos, allele);
}
}
else
{
sequence.Add(pos, allele);
}
}
long m_max = sequence.Keys.Max();
long m_min = sequence.Keys.Min();
StringBuilder sb = new StringBuilder();
for (long i = m_min; i <= m_max; i++)
{
if (sequence.ContainsKey(i))
{
sb.Append(sequence[i]);
}
else
{
sb.Append("N");
}
}
result.reliability = 100;
if (m_max - m_min != pos_end - pos_start)
{
result.reliability = (int)((m_max - m_min) * 100 / (pos_end - pos_start));
}
string seq = sb.ToString();
if (seq.Contains("N"))
{
result.reliability = (int)(((seq.Count() - seq.Count(x => x == 'N')) * result.reliability) / (m_max - m_min));
}
result.raw_seq = seq;
string html_seq = seq;
for (long i = pos_start; i < m_min; i++)
{
html_seq = "N" + html_seq;
}
for (long i = m_max; i <= pos_end; i++)
{
html_seq = html_seq + "N";
}
if (replace_regex == "CONCAT")
{
MatchCollection mc = Regex.Matches(seq, filter_regex);
seq = "";
foreach (Match m in mc)
{
seq += m.Value;
html_seq = html_seq.Replace(m.Value, Regex.Replace(m.Value, "(" + motif_regex + ")", "<span style='border-color:lightblue;border-style: solid; border-width: 1px;background: -webkit-linear-gradient(left top, white , lightblue); background: -o-linear-gradient(bottom right, white, lightblue); background: -moz-linear-gradient(bottom right, white, lightblue); background: linear-gradient(to bottom right, white , lightblue);color:blue'>$1</span>"));
}
}
else
{
html_seq = html_seq.Replace(seq, Regex.Replace(seq, "(" + motif_regex + ")", "<span style='border-color:lightblue;border-style: solid; border-width: 1px;background: -webkit-linear-gradient(left top, white , lightblue); background: -o-linear-gradient(bottom right, white, lightblue); background: -moz-linear-gradient(bottom right, white, lightblue); background: linear-gradient(to bottom right, white , lightblue);color:blue'>$1</span>"));
seq = Regex.Replace(seq, filter_regex, replace_regex);
}
result.html_seq = html_seq.Replace("N", "<span style='color:#a0a0a0;'>·</span>");
//Console.WriteLine(">> "+seq + "\r\n");
MatchCollection matches = Regex.Matches(seq, motif_regex);
result.value = matches.Count;
}
catch (Exception)
{
result.value = 0;
}
return(result);
}
