tools/lh3misc/seq/bc2rg.d

   1 #!/usr/bin/evn rdmd
   2
   3 import std.stdio, std.string, std.regex, klib;
   4
   5 void main(string[] args)
   6 {
   7         if (args.length < 4) {
   8                 writeln("Usage: rdmd bc2rg.d <sample> <barcode-list.txt> <aln.sam>");
   9                 return;
  10         }
  11         auto fp = new ZFile(args[2]);
  12         ubyte[] l;
  13         string[string] bc2rg;
  14         char[] base = ['A', 'C', 'G', 'T', 'N'];
  15         string[] rg;
  16         while (fp.readto(l) >= 0) {
  17                 auto t = std.string.split(cast(string)l);
  18                 if (t.length == 0) continue;
  19                 string r = args[1] ~ '-' ~ t[0] ~ '-' ~ t[1];
  20                 rg ~= r;
  21                 for (auto i = 0; i < t[0].length; ++i) {
  22                         foreach (c; base) {
  23                                 char[] s;
  24                                 s.length = t[0].length;
  25                                 s[] = cast(char[])t[0];
  26                                 s[i] = c;
  27                                 bc2rg[cast(string)s] = r;
  28                         }
  29                 }
  30         }
  31         fp.close();
  32         fp = new ZFile(args[3]);
  33         bool first = true;
  34         auto re = regex(r"^\S+#([ACGTN]+)");
  35         while (fp.readto(l) >= 0) {
  36                 if (l[0] != '@') {
  37                         if (first) {
  38                                 foreach (p; rg)
  39                                         writefln("@RG\tID:%s\tSM:%s\tLB:%s", p, args[1], p);
  40                                 first = false;
  41                         }
  42                         auto c = match(cast(string)l, re).captures;
  43                         if (c[1] in bc2rg) writeln(cast(char[])l, "\tRG:Z:", bc2rg[c[1]]);
  44                         else writeln(cast(char[])l);
  45                 } else writeln(cast(char[])l);
  46         }
  47 }