@datagrok/bio 1.3.1 → 1.4.2

package/detectors.js

@@ -11,7 +11,7 @@
 
 class BioPackageDetectors extends DG.Package {
 
-  static semType = 'MACROMOLECULE';
+  static mmSemType = 'Macromolecule';
 
   static Units = {
     FastaSeqPt: 'fasta:SEQ:PT', FastaSeqNt: 'fasta:SEQ:NT', FastaMsaPt: 'fasta:MSA:PT', FastaMsaNt: 'fasta:MSA:NT',
@@ -40,25 +40,33 @@ class BioPackageDetectors extends DG.Package {
     // TODO: Lazy calculations could be helpful for performance and convenient for expressing classification logic.
     const statsAsChars = BioPackageDetectors.getStats(col, 5, BioPackageDetectors.splitterAsChars);
     if (statsAsChars.sameLength) {
-      const alphabet = BioPackageDetectors.detectAlphabet(statsAsChars.freq, alphabetCandidates, '-');
-      const units = `fasta:SEQ.MSA:${alphabet}`;
-      col.setTag(DG.TAGS.UNITS, units);
-      return BioPackageDetectors.semType;
+      if (Object.keys(statsAsChars.freq).length > 0) { // require non empty alphabet
+        const alphabet = BioPackageDetectors.detectAlphabet(statsAsChars.freq, alphabetCandidates, '-');
+        const units = `fasta:SEQ.MSA:${alphabet}`;
+        col.setTag(DG.TAGS.UNITS, units);
+        return BioPackageDetectors.mmSemType;
+      }
     } else {
-      const sep = BioPackageDetectors.detectSeparator(statsAsChars.freq);
-      const gapSymbol = sep ? '' : '-';
-      const splitter = sep ? BioPackageDetectors.getSplitterWithSeparator(sep) : BioPackageDetectors.splitterAsFasta;
+      const separator = BioPackageDetectors.detectSeparator(statsAsChars.freq);
+      const gapSymbol = separator ? '' : '-';
+      const splitter = separator ? BioPackageDetectors.getSplitterWithSeparator(separator) : BioPackageDetectors.splitterAsFasta;
+
       const stats = BioPackageDetectors.getStats(col, 5, splitter);
+      if (Object.keys(stats.freq).length === 0) return null;
 
-      const format = sep ? 'separator' : 'fasta';
+      const format = separator ? 'separator' : 'fasta';
       const seqType = stats.sameLength ? 'SEQ.MSA' : 'SEQ';
 
       // TODO: If separator detected, then extra efforts to detect alphabet are allowed.
       const alphabet = BioPackageDetectors.detectAlphabet(stats.freq, alphabetCandidates, gapSymbol);
 
-      const units = `${format}:${seqType}:${alphabet}`;
-      col.setTag(DG.TAGS.UNITS, units);
-      return BioPackageDetectors.semType;
+      const forbidden = BioPackageDetectors.checkForbiddenWoSeparator(stats.freq);
+      if (separator || !forbidden) {
+        const units = `${format}:${seqType}:${alphabet}`;
+        col.setTag(DG.TAGS.UNITS, units);
+        if (separator) col.setTag('separator', separator);
+        return BioPackageDetectors.mmSemType;
+      }
     }
   }
 
@@ -73,9 +81,18 @@ class BioPackageDetectors extends DG.Package {
 
     // !!! But there is a caveat because exceptionally frequent char can be a gap symbol in MSA.
     // !!! What is the difference between the gap symbol and separator symbol in stats terms?
-
-    const maxFreq = Math.max(...Object.values(freq));
-    const sep = Object.entries(freq).find((kv) => kv[1] == maxFreq)[0];
+    // const noSeparatorRe = /[a-z\d]+$/i;
+    const noSeparatorRe = /[HBCNOFPSKVYI]/i; // Mendeleev's periodic table single char elements
+    const cleanFreq = Object.assign({}, ...Object.entries(freq)
+      .filter(([m, f]) => !noSeparatorRe.test(m) &&
+        !BioPackageDetectors.AminoacidsFastaAlphabet.has(m) &&
+        !BioPackageDetectors.NucleotidesFastaAlphabet.has(m))
+      .map(([m, f]) => ({[m]: f})));
+    if (Object.keys(cleanFreq).length == 0) return null;
+
+    const maxFreq = Math.max(...Object.values(cleanFreq));
+
+    const sep = Object.entries(freq).find(([k, v]) => v === maxFreq)[0];
     const sepFreq = freq[sep];
     const otherSumFreq = Object.entries(freq).filter((kv) => kv[0] !== sep)
       .map((kv) => kv[1]).reduce((pSum, a) => pSum + a, 0);
@@ -83,6 +100,12 @@ class BioPackageDetectors extends DG.Package {
     return sepFreq / otherSumFreq > freqThreshold ? sep : null;
   }
 
+  /** Without a separator, special symbols or digits are not allowed as monomers. */
+  static checkForbiddenWoSeparator(freq) {
+    const forbiddenRe = /[\d!@#$%^&*()_+\-=\[\]{};':"\\|,.<>\/?]/i;
+    return Object.keys(freq).filter((m) => forbiddenRe.test(m)).length > 0;
+  }
+
   /** Stats of sequences with specified splitter func, returns { freq, sameLength } */
   static getStats(seqCol, minLength, splitter) {
     const freq = {};