@teselagen/bio-parsers 0.4.9 → 0.4.10

package/genbankToJson.d.ts

@@ -1,3 +1,3 @@
-export function parseFeatureLocation(locStr: any, isProtein: any, inclusive1BasedStart: any, inclusive1BasedEnd: any): any[];
+export function parseFeatureLocation(locStr: any, isProtein: any, inclusive1BasedStart: any, inclusive1BasedEnd: any, isCircular: any, sequenceLength: any): any[];
 export default genbankToJson;
 declare function genbankToJson(string: any, options?: {}): any;

package/index.js

@@ -19772,7 +19772,7 @@ function flattenSequenceArray(parsingResultArray, opts) {
   return parsingResultArray;
 }
 __name(flattenSequenceArray, "flattenSequenceArray");
-function parseFeatureLocation(locStr, isProtein, inclusive1BasedStart, inclusive1BasedEnd) {
+function parseFeatureLocation(locStr, isProtein, inclusive1BasedStart, inclusive1BasedEnd, isCircular, sequenceLength) {
   locStr = locStr.trim();
   const locArr = [];
   locStr.replace(/(\d+)/g, function(string, match) {
@@ -19793,6 +19793,19 @@ function parseFeatureLocation(locStr, isProtein, inclusive1BasedStart, inclusive
       isProtein ? convertAACaretPositionOrRangeToDna(location) : location
     );
   }
+  if (isCircular) {
+    for (let i = 0; i < locArray.length; i += 2) {
+      const firstFeature = locArray[i];
+      const secondFeature = locArray[i + 1];
+      if (firstFeature.end === sequenceLength - (inclusive1BasedEnd ? 0 : 1) && secondFeature.start === 1 - (inclusive1BasedStart ? 0 : 1)) {
+        locArray[i] = {
+          start: firstFeature.start,
+          end: secondFeature.end
+        };
+        locArray.splice(i + 1, 1);
+      }
+    }
+  }
   return locArray;
 }
 __name(parseFeatureLocation, "parseFeatureLocation");
@@ -20126,7 +20139,9 @@ function genbankToJson(string, options = {}) {
             line.trim(),
             options.isProtein,
             inclusive1BasedStart,
-            inclusive1BasedEnd
+            inclusive1BasedEnd,
+            result.parsedSequence.circular,
+            result.parsedSequence.sequence.length
           )
         );
         lastLineWasLocation = true;
@@ -20159,7 +20174,9 @@ function genbankToJson(string, options = {}) {
             val2,
             options.isProtein,
             inclusive1BasedStart,
-            inclusive1BasedEnd
+            inclusive1BasedEnd,
+            result.parsedSequence.circular,
+            result.parsedSequence.sequence.length
           )
         );
         lastLineWasLocation = true;

package/index.mjs

@@ -19770,7 +19770,7 @@ function flattenSequenceArray(parsingResultArray, opts) {
   return parsingResultArray;
 }
 __name(flattenSequenceArray, "flattenSequenceArray");
-function parseFeatureLocation(locStr, isProtein, inclusive1BasedStart, inclusive1BasedEnd) {
+function parseFeatureLocation(locStr, isProtein, inclusive1BasedStart, inclusive1BasedEnd, isCircular, sequenceLength) {
   locStr = locStr.trim();
   const locArr = [];
   locStr.replace(/(\d+)/g, function(string, match) {
@@ -19791,6 +19791,19 @@ function parseFeatureLocation(locStr, isProtein, inclusive1BasedStart, inclusive
       isProtein ? convertAACaretPositionOrRangeToDna(location) : location
     );
   }
+  if (isCircular) {
+    for (let i = 0; i < locArray.length; i += 2) {
+      const firstFeature = locArray[i];
+      const secondFeature = locArray[i + 1];
+      if (firstFeature.end === sequenceLength - (inclusive1BasedEnd ? 0 : 1) && secondFeature.start === 1 - (inclusive1BasedStart ? 0 : 1)) {
+        locArray[i] = {
+          start: firstFeature.start,
+          end: secondFeature.end
+        };
+        locArray.splice(i + 1, 1);
+      }
+    }
+  }
   return locArray;
 }
 __name(parseFeatureLocation, "parseFeatureLocation");
@@ -20124,7 +20137,9 @@ function genbankToJson(string, options = {}) {
             line.trim(),
             options.isProtein,
             inclusive1BasedStart,
-            inclusive1BasedEnd
+            inclusive1BasedEnd,
+            result.parsedSequence.circular,
+            result.parsedSequence.sequence.length
           )
         );
         lastLineWasLocation = true;
@@ -20157,7 +20172,9 @@ function genbankToJson(string, options = {}) {
             val2,
             options.isProtein,
             inclusive1BasedStart,
-            inclusive1BasedEnd
+            inclusive1BasedEnd,
+            result.parsedSequence.circular,
+            result.parsedSequence.sequence.length
           )
         );
         lastLineWasLocation = true;

package/index.umd.js

@@ -19774,7 +19774,7 @@ var __async = (__this, __arguments, generator) => {
     return parsingResultArray;
   }
   __name(flattenSequenceArray, "flattenSequenceArray");
-  function parseFeatureLocation(locStr, isProtein, inclusive1BasedStart, inclusive1BasedEnd) {
+  function parseFeatureLocation(locStr, isProtein, inclusive1BasedStart, inclusive1BasedEnd, isCircular, sequenceLength) {
     locStr = locStr.trim();
     const locArr = [];
     locStr.replace(/(\d+)/g, function(string, match) {
@@ -19795,6 +19795,19 @@ var __async = (__this, __arguments, generator) => {
         isProtein ? convertAACaretPositionOrRangeToDna(location) : location
       );
     }
+    if (isCircular) {
+      for (let i2 = 0; i2 < locArray.length; i2 += 2) {
+        const firstFeature = locArray[i2];
+        const secondFeature = locArray[i2 + 1];
+        if (firstFeature.end === sequenceLength - (inclusive1BasedEnd ? 0 : 1) && secondFeature.start === 1 - (inclusive1BasedStart ? 0 : 1)) {
+          locArray[i2] = {
+            start: firstFeature.start,
+            end: secondFeature.end
+          };
+          locArray.splice(i2 + 1, 1);
+        }
+      }
+    }
     return locArray;
   }
   __name(parseFeatureLocation, "parseFeatureLocation");
@@ -20128,7 +20141,9 @@ var __async = (__this, __arguments, generator) => {
               line.trim(),
               options.isProtein,
               inclusive1BasedStart,
-              inclusive1BasedEnd
+              inclusive1BasedEnd,
+              result.parsedSequence.circular,
+              result.parsedSequence.sequence.length
             )
           );
           lastLineWasLocation = true;
@@ -20161,7 +20176,9 @@ var __async = (__this, __arguments, generator) => {
               val2,
               options.isProtein,
               inclusive1BasedStart,
-              inclusive1BasedEnd
+              inclusive1BasedEnd,
+              result.parsedSequence.circular,
+              result.parsedSequence.sequence.length
             )
           );
           lastLineWasLocation = true;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@teselagen/bio-parsers",
-  "version": "0.4.9",
+  "version": "0.4.10",
   "dependencies": {
     "@teselagen/sequence-utils": "0.3.15",
     "@teselagen/range-utils": "0.3.7",

package/src/genbankToJson.js

@@ -12,7 +12,9 @@ export function parseFeatureLocation(
   locStr,
   isProtein,
   inclusive1BasedStart,
-  inclusive1BasedEnd
+  inclusive1BasedEnd,
+  isCircular,
+  sequenceLength
 ) {
   locStr = locStr.trim();
   const locArr = [];
@@ -39,6 +41,30 @@ export function parseFeatureLocation(
       isProtein ? convertAACaretPositionOrRangeToDna(location) : location
     );
   }
+  // In genbank files, origin-spanning features are represented as follows:
+  // complement(join(490883..490885,1..879)) (for a circular sequence of length 490885)
+  // Then, for locations in locArray we check if there is a feature that ends at sequenceLength
+  // joined with a feature that starts at 1. If so, we merge them into a single feature.
+  // (see https://github.com/TeselaGen/tg-oss/issues/35)
+
+  if (isCircular) {
+    // Iterate by pairs of features
+    for (let i = 0; i < locArray.length; i += 2) {
+      const firstFeature = locArray[i];
+      const secondFeature = locArray[i + 1];
+      if (
+        firstFeature.end === sequenceLength - (inclusive1BasedEnd ? 0 : 1) &&
+        secondFeature.start === 1 - (inclusive1BasedStart ? 0 : 1)
+      ) {
+        // Merge the two features
+        locArray[i] = {
+          start: firstFeature.start,
+          end: secondFeature.end
+        };
+        locArray.splice(i + 1, 1);
+      }
+    }
+  }
   return locArray;
 }
 
@@ -468,7 +494,9 @@ function genbankToJson(string, options = {}) {
             line.trim(),
             options.isProtein,
             inclusive1BasedStart,
-            inclusive1BasedEnd
+            inclusive1BasedEnd,
+            result.parsedSequence.circular,
+            result.parsedSequence.sequence.length
           )
         );
         lastLineWasLocation = true;
@@ -513,7 +541,9 @@ function genbankToJson(string, options = {}) {
             val,
             options.isProtein,
             inclusive1BasedStart,
-            inclusive1BasedEnd
+            inclusive1BasedEnd,
+            result.parsedSequence.circular,
+            result.parsedSequence.sequence.length
           )
         );
         lastLineWasLocation = true;