Merge ebc41aa into a849211

README.md

@@ -24,7 +24,7 @@
 * **[VariantKey Format](#vkformat)**
     * [VariantKey Properties](#vkproperties)
 * [VariantKey Input values](#vkinput)
-* [RegionKey](#regionkey)
+* **[RegionKey](#regionkey)**
 * [Binary file formats for lookup tables](#binaryfiles)
 * [C Library](#clib)
 * [GO Library](#golib)
@@ -369,7 +369,11 @@ The VariantKey is composed of 3 sections arranged in 64 bit:
 <a name="regionkey"></a>
 ## RegionKey
 
-This library also includes functions to represent a human genetic region as number.
+*RegionKey* encodes a human genetic region (defined as the set of *chromosome*, *start position*, *end position* and *strand direction*) in a 64 bit unsigned integer number.
+
+RegionKey allows to repesent a region as a single entity, and provides analogous properties as the ones listed in [VariantKey Properties](#vkproperties).
+
+The encoding of the first 33 bit (CROM, STARTPOS) is the same as in VariantKey.
 
 The RegionKey is composed of 4 sections arranged in 64 bit:
 
@@ -433,7 +437,9 @@ The RegionKey is composed of 4 sections arranged in 64 bit:
     +1 : 1 dec = "01" bin
     ```
 
-The last bit of RegionKey is reserved.
+* The last bit of RegionKey is reserved.
+
+This software library provides several functions to operate with *RegionKey* and interact with *VariantKey*.
 
 
 <a name="binaryfiles"></a>

VERSION

@@ -1 +1 @@
-2.14.0
+2.15.0

c/src/nrvk.c

@@ -58,11 +58,9 @@ size_t find_ref_alt_by_variantkey(const unsigned char *src, uint64_t last, uint6
 
 size_t reverse_variantkey(const unsigned char *src, uint64_t last, uint64_t vk, variantkey_rev_t *rev)
 {
-    variantkey_t h = {0,0,0};
-    decode_variantkey(vk, &h);
-    decode_chrom(h.chrom, rev->chrom);
-    rev->pos = h.pos;
-    size_t len = decode_refalt(h.refalt, rev->ref, &rev->sizeref, rev->alt, &rev->sizealt);
+    decode_chrom(extract_variantkey_chrom(vk), rev->chrom);
+    rev->pos = extract_variantkey_pos(vk);
+    size_t len = decode_refalt(extract_variantkey_refalt(vk), rev->ref, &rev->sizeref, rev->alt, &rev->sizealt);
     if ((len == 0) && (last > 0))
     {
         len = find_ref_alt_by_variantkey(src, last, vk, rev->ref, &rev->sizeref, rev->alt, &rev->sizealt);

c/src/regionkey.c

@@ -80,12 +80,10 @@ void decode_regionkey(uint64_t code, regionkey_t *rk)
 
 void reverse_regionkey(uint64_t rk, regionkey_rev_t *rev)
 {
-    regionkey_t h = {0,0,0,0};
-    decode_regionkey(rk, &h);
-    decode_chrom(h.chrom, rev->chrom);
-    rev->startpos = h.startpos;
-    rev->endpos = h.endpos;
-    rev->strand = decode_region_strand(h.strand);
+    decode_chrom(extract_regionkey_chrom(rk), rev->chrom);
+    rev->startpos = extract_regionkey_startpos(rk);
+    rev->endpos = extract_regionkey_endpos(rk);
+    rev->strand = decode_region_strand(extract_regionkey_strand(rk));
 }
 
 uint64_t regionkey(const char *chrom, size_t sizechrom, uint32_t startpos, uint32_t endpos, int8_t strand)

conda/c.vk/meta.yaml

@@ -1,6 +1,6 @@
 package:
   name: vk
-  version: 2.14.0
+  version: 2.15.0
 
 source:
   path: ../..

conda/python/meta.yaml

@@ -1,6 +1,6 @@
 package:
   name: variantkey
-  version: 2.14.0
+  version: 2.15.0
 
 source:
   path: ../..

conda/r/meta.yaml

@@ -1,6 +1,6 @@
 package:
   name: r-variantkey
-  version: 2.14.0
+  version: 2.15.0
 
 source:
   path: ../..

javascript/Makefile

@@ -33,6 +33,7 @@ build:
 	@mkdir -p target/build
 	uglifyjs --compress --keep-fnames --comments --output target/build/variantkey.js src/variantkey.js 
 	cd test && node test_variantkey.js '../target/build/variantkey.js'
+	cd test && node test_regionkey.js '../target/build/variantkey.js'
 
 # Format the source code
 format:

javascript/src/variantkey.js

@@ -39,7 +39,7 @@ function encodeChrom(chrom) {
         return 0;
     }
     // X > 23 ; Y > 24 ; M > 25
-    onecharmap = [
+    var onecharmap = [
         0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
         0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
         /*                                     M                                 X   Y */
@@ -90,7 +90,7 @@ function encodeBase(c) {
       G > 2
       T > 3
     */
-    map = [
+    var map = [
         4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,
         4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,
         /* A     C           G                                      T */
@@ -259,7 +259,7 @@ function extractVariantKeyChrom(vk) {
 }
 
 function extractVariantKeyPos(vk) {
-    return (((vk.hi & 0x7FFFFFF) << 1) | (vk.lo >>> 31)) >>> 0;
+    return (((vk.hi & 0x07FFFFFF) << 1) | (vk.lo >>> 31)) >>> 0;
 }
 
 function extractVariantKeyRefAlt(vk) {
@@ -274,6 +274,16 @@ function decodeVariantKey(vk) {
     };
 }
 
+function reverseVariantKey(vk) {
+    var ra = decodeRefAlt(extractVariantKeyRefAlt(vk));
+    return {
+        "chrom": decodeChrom(extractVariantKeyChrom(vk)),
+        "pos": extractVariantKeyPos(vk),
+        "ref": ra.ref,
+        "alt": ra.alt
+    }
+}
+
 function variantKey(chrom, pos, ref, alt) {
     return encodeVariantKey(encodeChrom(chrom), pos, encodeRefAlt(ref, alt));
 }
@@ -322,14 +332,90 @@ function parseHex(vs) {
     };
 }
 
-function reverseVariantKey(dvk) {
-    var ra = decodeRefAlt(dvk.refalt);
+function encodeRegionStrand(strand) {
+    var map = [2, 0, 1, 0];
+    return map[((++strand) & 3)];
+}
+
+function decodeRegionStrand(strand) {
+    var map = [0, 1, -1, 0];
+    return map[(strand & 3)];
+}
+
+function encodeRegionKey(chrom, startpos, endpos, strand) {
     return {
-        "chrom": decodeChrom(dvk.chrom),
-        "pos": dvk.pos,
-        "ref": ra.ref,
-        "alt": ra.alt
-    }
+        "hi": ((((chrom >>> 0) << 27) | (startpos >>> 1)) >>> 0),
+        "lo": ((((startpos >>> 0) << 31) | ((endpos >>> 0) << 3) | ((strand >>> 0) << 1)) >>> 0)
+    };
+}
+
+function extractRegionKeyChrom(rk) {
+    return ((rk.hi & 0xF8000000) >>> 27);
+}
+
+function extractRegionKeyStartPos(rk) {
+    return (((rk.hi & 0x07FFFFFF) << 1) | (rk.lo >>> 31)) >>> 0;
+}
+
+function extractRegionKeyEndPos(rk) {
+    return (rk.lo & 0x7FFFFFF8) >>> 3
+}
+
+function extractRegionKeyStrand(rk) {
+    return (rk.lo & 0x00000006) >>> 1;
+}
+
+function decodeRegionKey(rk) {
+    return {
+        "chrom": extractRegionKeyChrom(rk),
+        "startpos": extractRegionKeyStartPos(rk),
+        "endpos": extractRegionKeyEndPos(rk),
+        "strand": extractRegionKeyStrand(rk)
+    };
+}
+
+function reverseRegionKey(rk) {
+    return {
+        "chrom": decodeChrom(extractRegionKeyChrom(rk)),
+        "startpos": extractRegionKeyStartPos(rk),
+        "endpos": extractRegionKeyEndPos(rk),
+        "strand": decodeRegionStrand(extractRegionKeyStrand(rk))
+    };
+}
+
+function regionKey(chrom, startpos, endpos, strand) {
+    return encodeRegionKey(encodeChrom(chrom), startpos, endpos, encodeRegionStrand(strand));
+}
+
+function regionKeyString(rk) {
+    return padL08(rk.hi.toString(16)) + padL08(rk.lo.toString(16));
+}
+
+function getVariantKeyEndPos(vk) {
+    return extractVariantKeyPos(vk) + ((vk.lo & 0x78000000) >>> 27);
+}
+
+function areOverlappingRegions(a_chrom, a_startpos, a_endpos, b_chrom, b_startpos, b_endpos) {
+    return ((a_chrom == b_chrom) && (a_startpos < b_endpos) && (a_endpos > b_startpos));
+}
+
+function areOverlappingRegionRegionKey(chrom, startpos, endpos, rk) {
+    return ((chrom == extractRegionKeyChrom(rk)) && (startpos < extractRegionKeyEndPos(rk)) && (endpos > extractRegionKeyStartPos(rk)));
+}
+
+function areOverlappingRegionKeys(rka, rkb) {
+    return ((extractRegionKeyChrom(rka) == extractRegionKeyChrom(rkb)) && (extractRegionKeyStartPos(rka) < extractRegionKeyEndPos(rkb)) && (extractRegionKeyEndPos(rka) > extractRegionKeyStartPos(rkb)));
+}
+
+function areOverlappingVariantKeyRegionKey(vk, rk) {
+    return ((extractVariantKeyChrom(vk) == extractRegionKeyChrom(rk)) && (extractVariantKeyPos(vk) < extractRegionKeyEndPos(rk)) && (getVariantKeyEndPos(vk) > extractRegionKeyStartPos(rk)));
+}
+
+function variantKeyToRegionKey(vk) {
+    return {
+        "hi": vk.hi,
+        "lo": ((vk.lo & 0x80000000) | ((getVariantKeyEndPos(vk) << 3) >>> 0)) >>> 0
+    };
 }
 
 if (typeof(module) !== 'undefined') {
@@ -351,5 +437,22 @@ if (typeof(module) !== 'undefined') {
         compareVariantKeyChromPos: compareVariantKeyChromPos,
         variantKeyString: variantKeyString,
         reverseVariantKey: reverseVariantKey,
+        encodeRegionStrand: encodeRegionStrand,
+        decodeRegionStrand: decodeRegionStrand,
+        encodeRegionKey: encodeRegionKey,
+        extractRegionKeyChrom: extractRegionKeyChrom,
+        extractRegionKeyStartPos: extractRegionKeyStartPos,
+        extractRegionKeyEndPos: extractRegionKeyEndPos,
+        extractRegionKeyStrand: extractRegionKeyStrand,
+        decodeRegionKey: decodeRegionKey,
+        reverseRegionKey: reverseRegionKey,
+        regionKey: regionKey,
+        regionKeyString: regionKeyString,
+        getVariantKeyEndPos: getVariantKeyEndPos,
+        areOverlappingRegions: areOverlappingRegions,
+        areOverlappingRegionRegionKey: areOverlappingRegionRegionKey,
+        areOverlappingRegionKeys: areOverlappingRegionKeys,
+        areOverlappingVariantKeyRegionKey: areOverlappingVariantKeyRegionKey,
+        variantKeyToRegionKey: variantKeyToRegionKey,
     }
 }