Add functions to retrieve all rsIDs from a VariantKey

VERSION

@@ -1 +1 @@
-5.3.5
+5.4.0

c/doc/Doxyfile

@@ -32,7 +32,7 @@ PROJECT_NAME           = "VariantKey"
 # This could be handy for archiving the generated documentation or
 # if some version control system is used.
 
-PROJECT_NUMBER = 5.3.5
+PROJECT_NUMBER = 5.4.0
 
 # Using the PROJECT_BRIEF tag one can provide an optional one line description
 # for a project that appears at the top of each page and should give viewer

c/src/variantkey/rsidvar.h

@@ -167,6 +167,27 @@ static inline uint32_t find_vr_rsid_by_variantkey(rsidvar_cols_t cvr, uint64_t *
     return *(cvr.rs + found);
 }
 
+/**
+ * Get the next rsID for the specified VariantKey in the VR file.
+ * This function should be used after find_vr_rsid_by_variantkey.
+ * This function can be called in a loop to get all rsIDs that are associated with the same VariantKey (if any).
+ *
+ * @param cvr       Structure containing the pointers to the VKRS memory mapped file columns (vkrs.bin).
+ * @param pos       Pointer to the current item. This will hold the position of the next record.
+ * @param last      Element (up to but not including) where to end the search (max value = nitems).
+ * @param vk        VariantKey.
+ *
+ * @return rsID data or zero data if not found
+ */
+static inline uint32_t get_next_vr_rsid_by_variantkey(rsidvar_cols_t cvr, uint64_t *pos, uint64_t last, uint64_t vk)
+{
+    if (col_has_next_uint64_t(cvr.vk, pos, last, vk))
+    {
+        return *(cvr.rs + *pos);
+    }
+    return 0;
+}
+
 /**
  * Search for the specified CHROM-POS range and returns the first occurrence of rsID in the VR file.
  *

c/test/test_example.c

@@ -283,9 +283,14 @@ int main()
     first = 0;
     uint64_t last = 9;
     rsid = find_vr_chrompos_range(cvr, &first, &last, 0x14, 0x000256C5, 0x000256CB);
-    fprintf(stdout, "%" PRIu32 "%" PRIu64 "%" PRIu64 "\n", rsid, first, last);
+    fprintf(stdout, "%" PRIu32 " %" PRIu64 " %" PRIu64 "\n", rsid, first, last);
     // 9973 7 8
 
+    fpos = 2;
+    rsid = get_next_vr_rsid_by_variantkey(cvr, &fpos, 9, 0x80010274003A0000);
+    fprintf(stdout, "%" PRIu32 " %" PRIu64 "\n", rsid, fpos);
+    // 97 3
+
     err = munmap_binfile(vr);
     if (err != 0)
     {

c/test/test_rsidvar.c

@@ -195,6 +195,35 @@ int test_find_vr_rsid_by_variantkey_notfound(rsidvar_cols_t cvr)
     return errors;
 }
 
+int test_get_next_vr_rsid_by_variantkey(rsidvar_cols_t cvr)
+{
+    int errors = 0;
+    uint64_t pos = 2;
+    uint32_t rsid = get_next_vr_rsid_by_variantkey(cvr, &pos, cvr.nrows, 0x80010274003A0000);
+    if (pos != 3)
+    {
+        fprintf(stderr, "%s (1 Expected) pos 3, got %" PRIu64 "\n", __func__, pos);
+        ++errors;
+    }
+    if (rsid != 97)
+    {
+        fprintf(stderr, "%s (1) Expected rsid 97, got %" PRIx32 "\n", __func__, rsid);
+        ++errors;
+    }
+    rsid = get_next_vr_rsid_by_variantkey(cvr, &pos, cvr.nrows, 0x80010274003A0000);
+    if (pos != 4)
+    {
+        fprintf(stderr, "%s (2) Expected pos 4, got %" PRIu64 "\n", __func__, pos);
+        ++errors;
+    }
+    if (rsid != 0)
+    {
+        fprintf(stderr, "%s (2) Expected rsid 0, got %" PRIx32 "\n", __func__, rsid);
+        ++errors;
+    }
+    return errors;
+}
+
 int test_find_vr_chrompos_range(rsidvar_cols_t cvr)
 {
     int errors = 0;
@@ -340,6 +369,7 @@ int main()
     errors += test_get_next_rv_variantkey_by_rsid(crv);
     errors += test_find_vr_rsid_by_variantkey(cvr);
     errors += test_find_vr_rsid_by_variantkey_notfound(cvr);
+    errors += test_get_next_vr_rsid_by_variantkey(cvr);
     errors += test_find_vr_chrompos_range(cvr);
     errors += test_find_vr_chrompos_range_notfound(cvr);
 

conda/c.src/meta.yaml

@@ -1,6 +1,6 @@
 package:
   name: variantkey-src
-  version: 5.3.5
+  version: 5.4.0
 
 source:
   path: ../..

conda/c.vk/meta.yaml

@@ -1,6 +1,6 @@
 package:
   name: variantkey-vk
-  version: 5.3.5
+  version: 5.4.0
 
 source:
   path: ../..

conda/python-class/meta.yaml

@@ -1,6 +1,6 @@
 package:
   name: pyvariantkey
-  version: 5.3.5
+  version: 5.4.0
 
 source:
   path: ../..
@@ -14,11 +14,11 @@ requirements:
     - setuptools
     - numpy >=1.15.0
   build:
-    - variantkey >=5.3.5
+    - variantkey >=5.4.0
     - numpy >=1.15.0
   run:
     - python
-    - variantkey >=5.3.5
+    - variantkey >=5.4.0
     - numpy >=1.15.0
 
 test:
@@ -30,7 +30,7 @@ test:
     - pytest-cov
     - pytest-benchmark
     - pycodestyle
-    - variantkey >=5.3.5
+    - variantkey >=5.4.0
     - numpy >=1.15.0
   imports:
     - pyvariantkey

conda/python/meta.yaml

@@ -1,6 +1,6 @@
 package:
   name: variantkey
-  version: 5.3.5
+  version: 5.4.0
 
 source:
   path: ../..

conda/r/meta.yaml

@@ -1,6 +1,6 @@
 package:
   name: r-variantkey
-  version: 5.3.5
+  version: 5.4.0
 
 source:
   path: ../..

go/example/main.go

@@ -178,8 +178,14 @@ func main() {
 	fmt.Println(vr.FindVRRsidByVariantKey(0, 9, 0x80010274003A0000))
 	// 97 3
 
+	fmt.Println(vr.GetNextVRRsidByVariantKey(2, 9, 0x80010274003A0000))
+	// 97 3
+
+	fmt.Println(vr.FindAllVRRsidByVariantKey(0, 9, 0x80010274003A0000))
+	// [97]
+
 	fmt.Println(vr.FindVRChromPosRange(0, 9, 0x14, 0x000256C5, 0x000256CB))
-	// 9973 7 8
+	// 9973 7 9
 
 	// /\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\
 

go/src/rsidvar_test.go

@@ -138,6 +138,35 @@ func BenchmarkFindVRRsidByVariantKey(b *testing.B) {
 	}
 }
 
+func TestGetNextVRRsidByVariantKey(t *testing.T) {
+	var rsid uint32
+	pos := uint64(2)
+	rsid, pos = vr.GetNextVRRsidByVariantKey(pos, vr.NRows, 0x80010274003A0000)
+	if pos != 3 {
+		t.Errorf("(1) Expected pos 3, got %d", pos)
+	}
+	if rsid != 97 {
+		t.Errorf("(1) Expected rsID 97, got %d", rsid)
+	}
+	rsid, pos = vr.GetNextVRRsidByVariantKey(pos, vr.NRows, 0x80010274003A0000)
+	if pos != 4 {
+		t.Errorf("(2) Expected pos 4, got %d", pos)
+	}
+	if rsid != 0 {
+		t.Errorf("(2) Expected rsID 0, got %d", rsid)
+	}
+}
+
+func TestFindAllVRRsidByVariantKey(t *testing.T) {
+	rsid := vr.FindAllVRRsidByVariantKey(0, vr.NRows, 0x80010274003A0000)
+	if len(rsid) != 1 {
+		t.Errorf("Expected len 1, got %d", len(rsid))
+	}
+	if rsid[0] != 97 {
+		t.Errorf("Expected rsID 97, got %d", rsid[0])
+	}
+}
+
 func TestFindVRChromPosRange(t *testing.T) {
 	rsid, first, last := vr.FindVRChromPosRange(0, vr.NRows, testData[6].chrom, testData[7].pos, testData[8].pos)
 	if rsid != testData[7].rsid {

go/src/variantkey.go

@@ -407,6 +407,29 @@ func (crv RSIDVARCols) FindVRRsidByVariantKey(first uint64, last uint64, vk uint
 	return rsid, uint64(cfirst)
 }
 
+// GetNextVRRsidByVariantKey get the next rsID for the specified VariantKey in the VR file.
+// Returns the rsID or 0, and the position
+func (cvr RSIDVARCols) GetNextVRRsidByVariantKey(pos, last uint64, vk uint64) (uint32, uint64) {
+	cpos := C.uint64_t(pos)
+	rsid := uint32(C.get_next_vr_rsid_by_variantkey(castGoRSIDVARColsToC(cvr), &cpos, C.uint64_t(last), C.uint64_t(vk)))
+	return rsid, uint64(cpos)
+}
+
+// FindAllVRRsidByVariantKey get all rsID for the specified VariantKeys in the VR file.
+// Returns a list of rsIDs
+func (cvr RSIDVARCols) FindAllVRRsidByVariantKey(first, last uint64, vk uint64) (rsids []uint32) {
+	ccr := castGoRSIDVARColsToC(cvr)
+	cfirst := C.uint64_t(first)
+	clast := C.uint64_t(last)
+	cvk := C.uint64_t(vk)
+	rsid := uint32(C.find_vr_rsid_by_variantkey(ccr, &cfirst, clast, cvk))
+	for rsid > 0 {
+		rsids = append(rsids, rsid)
+		rsid = uint32(C.get_next_vr_rsid_by_variantkey(ccr, &cfirst, clast, cvk))
+	}
+	return
+}
+
 // FindVRChromPosRange search for the specified CHROM-POS range and returns the first occurrence of RSID in the VR file.
 func (crv RSIDVARCols) FindVRChromPosRange(first, last uint64, chrom uint8, posMin, posMax uint32) (uint32, uint64, uint64) {
 	cfirst := C.uint64_t(first)

python-class/.gitignore

@@ -1,5 +1,5 @@
 .cache
-.coverage*
+.coverage
 LICENSE
 README.md
 c/

python-class/pyvariantkey/variantkey.py

@@ -432,7 +432,7 @@ def find_all_rv_variantkey_by_rsid(self, rsid):
         vk = []
         rsid_arr = np.array(rsid).astype(np.uint32)
         for x in np.nditer(rsid_arr):
-            vk = vk + pvk.find_all_rv_variantkey_by_rsid(self.rsvk_mc, 0, self.rsvk_nrows, x)
+            vk = vk + pvk.find_all_rv_variantkey_by_rsid(self.rsvk_mc, 0, self.rsvk_nrows, x.item(0))
         return np.array(vk).astype(np.uint64)
 
     def find_vr_rsid_by_variantkey(self, vk):
@@ -457,6 +457,49 @@ def find_vr_rsid_by_variantkey(self, vk):
                  self.vkrs_nrows,
                  np.array(vk).astype(np.uint64))
 
+    def get_next_vr_rsid_by_variantkey(self, pos, vk):
+        """Get the next rsID for the specified VariantKey in the VR file."\
+        " This function should be used after find_vr_rsid_by_variantkey."\
+        " This function can be called in a loop to get all rsIDs that are associated with the same VariantKey (if any).
+
+        Parameters
+        ----------
+        pos : uint64
+            Current item position.
+        vk : uint64
+            variantKey to search.
+
+        Returns
+        -------
+        tuple :
+            - uint32 : rsID or 0 in case not found.
+            - uint64 : Item position in the file.
+        """
+        f = np.vectorize(pvk.get_next_vr_rsid_by_variantkey, excluded=['mc', 'last'], otypes=[np.uint32, np.uint64])
+        return f(self.vkrs_mc,
+                 np.array(pos).astype(np.uint64),
+                 self.vkrs_nrows,
+                 np.array(vk).astype(np.uint64))
+
+    def find_all_vr_rsid_by_variantkey(self, vk):
+        """Search for the specified VariantKey and returns all associated rsIDs.
+
+        Parameters
+        ----------
+        vk : uint64
+            variantKey to search.
+
+        Returns
+        -------
+        uint32 :
+            - rsID(s).
+        """
+        rs = []
+        vk_arr = np.array(vk).astype(np.uint64)
+        for x in np.nditer(vk_arr):
+            rs = rs + pvk.find_all_vr_rsid_by_variantkey(self.vkrs_mc, 0, self.vkrs_nrows, x.item(0))
+        return np.array(rs).astype(np.uint32)
+
     def find_vr_chrompos_range(self, chrom, pos_min, pos_max):
         """Search for the specified CHROM-POS range and returns the first occurrence of rsID in the VR file.
 

python-class/setup.py

@@ -33,7 +33,7 @@ def run(self):
 
 setup(
     name='pyvariantkey',
-    version='5.3.5.1',
+    version='5.4.0.1',
     keywords=('variantkey variant key genetic genomics'),
     description="VariantKey Python wrapper class",
     long_description=read('../README.md'),
@@ -51,7 +51,7 @@ def run(self):
     ],
     install_requires=[
         'numpy>=1.15.0',
-        'variantkey>=5.3.5.1',
+        'variantkey>=5.4.0.1',
     ],
     extras_require={
         'test': [

python-class/test/example.py

@@ -149,6 +149,12 @@
 vk.find_vr_rsid_by_variantkey(vk=0x80010274003a0000)
 # (array(97, dtype=uint32), array(3, dtype=uint64))
 
+vk.get_next_vr_rsid_by_variantkey(2, vk=0x80010274003a0000)
+# (array(97, dtype=uint32), array(3, dtype=uint64))
+
+vk.find_all_vr_rsid_by_variantkey(0x80010274003a0000)
+# array([97], dtype=uint32)
+
 vk.find_vr_chrompos_range(0x14, 0x000256c5, 0x000256cb)
 # (array(9973, dtype=uint32), array(7, dtype=uint64), array(9, dtype=uint64))
 

python-class/test/test_rsidvar.py

@@ -58,10 +58,10 @@ def test_find_rv_variantkey_by_rsid_notfound(self):
         np.testing.assert_array_equal(first, 9)
 
     def test_get_next_rv_variantkey_by_rsid(self):
-        vk, pos = npvk.get_next_rv_variantkey_by_rsid(2, 0x00000061)
-        np.testing.assert_array_equal(vk, 0x80010274003A0000)
+        vk, pos = npvk.get_next_rv_variantkey_by_rsid(2, testData[3, 4].astype(np.uint32))
+        np.testing.assert_array_equal(vk, testData[3, 5].astype(np.uint64))
         np.testing.assert_array_equal(pos, 3)
-        vk, pos = npvk.get_next_rv_variantkey_by_rsid(pos, 0x00000061)
+        vk, pos = npvk.get_next_rv_variantkey_by_rsid(pos, testData[3, 4].astype(np.uint32))
         np.testing.assert_array_equal(vk, 0)
         np.testing.assert_array_equal(pos, 4)
 
@@ -79,6 +79,18 @@ def test_find_vr_rsid_by_variantkey_notfound(self):
         np.testing.assert_array_equal(rx, 0)
         np.testing.assert_array_equal(first, 9)
 
+    def test_get_next_vr_rsid_by_variantkey(self):
+        rsid, pos = npvk.get_next_vr_rsid_by_variantkey(2, testData[3, 5].astype(np.uint64))
+        np.testing.assert_array_equal(rsid, 97)
+        np.testing.assert_array_equal(pos, 3)
+        rsid, pos = npvk.get_next_vr_rsid_by_variantkey(pos, testData[3, 5].astype(np.uint64))
+        np.testing.assert_array_equal(rsid, 0)
+        np.testing.assert_array_equal(pos, 4)
+
+    def test_find_all_vr_rsid_by_variantkey(self):
+        rsids = npvk.find_all_vr_rsid_by_variantkey(testData[3, 5].astype(np.uint64))
+        np.testing.assert_array_equal(len(rsids), 1)
+
     def test_find_vr_chrompos_range(self):
         xrsid, xfirst, xlast = npvk.find_vr_chrompos_range(0x14, 0x000256c5, 0x000256cb)
         np.testing.assert_array_equal(xrsid, 0x000026f5)

python/setup.py

@@ -30,7 +30,7 @@ def run(self):
 
 setup(
     name='variantkey',
-    version='5.3.5.1',
+    version='5.4.0.1',
     keywords=('variantkey variant key genetic genomics'),
     description="VariantKey Bindings for Python",
     long_description=read('../README.md'),

python/test/example.py

@@ -185,6 +185,12 @@
 vk.find_vr_rsid_by_variantkey(mc, 0, nrows, vk=0X80010274003A0000)
 # (97, 3)
 
+vk.get_next_vr_rsid_by_variantkey(mc, 2, nrows, vk=0X80010274003A0000)
+# (97, 3)
+
+vk.find_all_vr_rsid_by_variantkey(mc, 0, nrows, vk=0X80010274003A0000)
+# [97]
+
 vk.find_vr_chrompos_range(mc, 0, nrows, 0X14, 0X000256C5, 0X000256CB)
 # (9973, 7, 9)