biojs-io-clustal/lib/index.js

Parses clustal files

// Generated by CoffeeScript 1.8.0
var Clustal, GenericReader, st;

GenericReader = require("biojs-io-parser");

st = require("msa-seqtools");

module.exports = Clustal = {
  parse: function(text) {
    var blockstate, cSeq, k, keys, label, line, lines, match, obj, regex, seqCounter, seqs, sequence;
    seqs = [];
    if (Object.prototype.toString.call(text) === '[object Array]') {
      lines = text;
    } else {
      lines = text.split("\n");
    }
    if (lines[0].slice(0, 6) === !"CLUSTAL") {
      throw new Error("Invalid CLUSTAL Header");
    }
    k = 0;
    blockstate = 1;
    seqCounter = 0;
    while (k < lines.length) {
      k++;
      line = lines[k];
      if ((line == null) || line.length === 0) {
        blockstate = 1;
        continue;
      }
      if (line.trim().length === 0) {
        blockstate = 1;
        continue;
      } else {
        if (st.contains(line, "*")) {
          continue;
        }
        if (blockstate === 1) {
          seqCounter = 0;
          blockstate = 0;
        }
        regex = /^(?:\s*)(\S+)(?:\s+)(\S+)(?:\s*)(\d*)(?:\s*|$)/g;
        match = regex.exec(line);
        if (match != null) {
          label = match[1];
          sequence = match[2];
          if (seqCounter >= seqs.length) {
            obj = st.getMeta(label);
            label = obj.name;
            cSeq = new st.model(sequence, label, seqCounter);
            cSeq.ids = obj.ids || {};
            cSeq.details = obj.details || {};
            keys = Object.keys(cSeq.ids);
            if (keys.length > 0) {
              cSeq.id = cSeq.ids[keys[0]];
            }
            seqs.push(cSeq);
          } else {
            seqs[seqCounter].seq += sequence;
          }
          seqCounter++;
        } else {
          console.log("parse error", line);
        }
      }
    }
    return seqs;
  }
};

GenericReader.mixin(Clustal);